Question regarding permanent tiger I losses...

[JasonC]

MikeyD said "number of kills versus number of kills, not number of vehicles vs number of vehicles." And you think "number of kills" is well defined. It just isn't. I'll explain what I mean.

A Tiger is hit by a 76mm AP shell in the right front spocket wheel. This wrecks the track and immobilizes the Tiger. It survives the fight and is recovered. It spends a month in repair. A month later it is involved in a large retreat. The repaired drive fails during a road march. The tank has to be abandoned because the Russians are close, and it blown up by the crew. Not a kill? What role does the 76mm AP hit play in the eventual loss?

Another Tiger is holed 3 times by 76mm APCR, from the side. Most of the crew is killed, but it does not burn. The wreck is recovered and carried on the rolls as "long term repair". It is stripped of parts to repair other damaged Tigers. When the front eventually moves, it is not considered worth moving (having already been stripped, and extensively damaged inside by the immediate hits), and is blown up by the crew. A kill? What role did the 76mm APCR hits play in the eventual loss?

Now, what difference is there is how the two will be accounted for in loss statistics or reported tank strength? Both will be out of "runners" as soon as hit. Both will be in repair for an interim period. Both will eventually by TWO, but not listed as lost in combat, instead listed as "abandoned" or "blown up by crew". We know essentially nothing about internal realities on this subjects.

All vehicles, on the other hand, eventually end up in two categories. TWO - KOed - no longer around, or survived the war. Yes, some are lost by roundabout processes like the first above. Some are lost without taking battle damage - though that is quite rare, since anything lightly damaged is usually recovered for repair, and circumstances of loss typically require something to be wrong with the vehicle - with battle damage recent or older a leading cause of that. But this happens on both sides. So not every Tiger lost was KOed in combat - fine, not every western AFV lost was KOed in combat either.

If the ratios are slightly different, you can nudge estimates around slightly. What you can't do is practice "one entry accounting", with 25% of Tigers killing 165% of the Allied AFVs available to be killed by them. Which is how far you'd have to stretch to reach a figure like 33 each, when 4-5 each is all the overall numbers can accomodate.

[JasonC]

Because there is so much confusion over these sorts of attrition accounting issues, I state some theorems about attrition and weapon system effectiveness. They are truisms, but rarely reflected on, especially by those looking at the matter tactically and then expecting things to scale up linearly.

Theorem - the average major weapon system does not manage to take out a single comparable enemy weapon system over its entire combat lifetime.

Proof - the total number of weapon systems deployed is enemy plus friendly - call this a. Somebody wins the war. Some of his weapons systems survive the war - call this b. The total number of weapon systems destroyed is enemy plus friendly minus survived - a minus b. This is less than the previous, because b is positive. The average weapon systems destroyed is total destroyed divided by total doing the destroying. This is a minus b, b greater than zero, divided by a. Which is less than one.

Theorem - the achieved average kill chance per shot (lethality adjusted "accuracy") of major weapon systems is typically less than one divided by total ammo load for that weapon system.

Proof - from the previous theorem, the average weapon system doesn't kill its own opposite number even once over its entire combat lifetime. From the existence of resupply issues, the average weapon fires off its whole ammo load. (Some die before doing so, many live long enough to be resupplied many times). Kills are therefore less than shots by at least the ammo load ratio.

Corollary (example) - in typical premodern battles where the losses on both sides were essentially always under 1/3rd of the men who walked onto the field, and the average ammo load per man who walked onto the field was 60-75 rounds, the achieved accuracy was on the order of 1% (to within a factor of 2 either way). Even though some of the shooting was at massed infantry targets, sometimes at close range.

Explanation - blind fire and extreme range fire accounts for a large portion of ammo expenditure. The average *kill* may happen at medium to short range. But the average *shot* is almost always taken at long range, against barely visible or unseen targets, etc.

Theorem - Rate of fire never scales, and is rarely of fundamental importance for weapon effectiveness.

Proof - The average weapon lives long enough to fire multiples of its ammo load. No army in history has ever been supplied with ammo as fast as it could throw it. Practical ammo expenditure limits over the medium term are logistical, not weapon determined. A higher rate of fire may allow shots to concentrate in narrow windows of time during which the enemy is more than usually exposed, and so be beneficial. But such effects will not scale, since they only help by exploiting far from average levels of enemy exposure. Sustained attempted use of higher ROF automatically hits logistical limits.

Moreover, it is not the case that average fire is KOing comparable enemy weapons in short time windows, making it critical to reducing his replies - because average fire does not succeed in destroying anything comparable over the whole combat life of the weapon system. It therefore is not, typically, doing so multiple times over much shorter time scales. There may be exceptional cases (outliers) where this happens and matters. But it cannot be the rule.

Theorem - a weapon system that typically manages to destroy its own number or more of the enemy and survive is an above average, successful weapon system.

Proof - follows immediately from the first theorem. All weapons cannot be above average. Any weapon that KOs its own counterpart or more is above average for weapon systems fielded by all sides combined. Any weapon system that does so and survives might well be able to do so again.

Theorem - The primary determinant of the lifetime effectiveness of a major weapon system is its survivability.

Proof - Essentially all weapon systems are designed to be able to take out something. All are regularly provided with vastly more ammo than they need to destroy a comparable enemy force under favorable conditions. The average system does not manage to destroy even one - for lack of targets, tactical conditions that allow its use, or because it has been KOed itself before managing to do so. All of the other factors except being KOed itself can be remedied by being in action longer, if there is any sense in the design. The only thing that will permanently keep a typical weapon from ever taking out something comparable, is being KOed first itself.

Wars last a long time, long enough to expend large amounts of ammo per weapon. Longevity therefore directly multiplies lifetime effectiveness.

Corollary - survivability is readily achievable for major weapon systems, by tactical employment if not by design characteristics.

Proof - wars take a long time, with large forces continually deployed, but succeed in destroying fewer weapon systems than are deployed overall. Achieved lethality to each fielded weapon per unit time is therefore necessarily low. Tactically, just staying out of areas of enemy dominance clearly increases the survivability of any system. Naturally the enemy tries to arrange the reverse. But it takes the whole length of the war for the winning side to successfully do so, for the whole losing side weapon set.

Now, none of these have anything to do with Tigers. But they do directly address the idea that a typical successful weapon system - to be successful - must destroy gobs of comparable enemy systems regularly, must achieve high per shot kill rates, etc. It is not so. By almost mathematical law.

Some subset of far above average weapon systems can account for more than their share of enemy systems, not truly comparable but in the same rough category. Most cannot. Plain vanilla ones cannot. The side that loses fewer systems may have better KO ratios than the overall average. But his average system cannot have a ratio better than the total systems the enemy deploys divided by the total systems he deploys against them. And the overall average for both sides must be lower still.

T-34s and Shermans cannot have accounted for a tank apiece. Trivially, since there were far more T-34s and Shermans than all Axis tanks combined. They couldn't have even accounted for a PAK apiece, for the same reason.

They might have accounted for a few infantrymen apiece, certainly - but not for 150 (that would be all German casualties) or realistically even 50 (because arty did most of it, small arms some of the remainder, etc).

Not in one 30 minute firefight, over their entire operational life. Yet they were highly effective weapon systems that played a leading role in actually winning the war. A weapon system that takes out its own number or more is above average and successful. But a weapon system need not accomplish this to be a success. It is a sufficient, not a necessary, criterion.

For what it is worth.

[ May 05, 2004, 05:31 PM: Message edited by: JasonC ]

[Mr. Tittles]

Ok. Lets say you have a Soviet 76mm ATG. It KOs two panzers but is captured. It is then used to destroy 4 T34 and a small horse drawn wagon. How do we do the accounting on this gun?

[JasonC]

You count it as 1 Russian weapons system with 2 AFVs KOed, and 1 German weapon system with 4 AFVs KOed. Simple. You count captures as a suppliment to German armaments production.

Also, you notice that this cannot be an average performance by a Russian 76mm gun, because there aren't twice as many German tanks produced as Russian 76mm guns produced. In fact, Russian AFVs might have averaged 0.25 kills apiece, and their guns even less than that.

[JasonC]

Yes Mr. Tittles, and when you read such things you should take another step beyond "amazed" to "incredulous" and then you should examine the matter, and you will find that either (1) it flat isn't so or (2) it is a rare outlier and completely unrepresentative of the average result achieved by StuGs.

The average StuG did not KO 8 tanks. That would require 67,000 tanks KOed by long StuGs alone, which are only 1/6th of the German AFV fleet, not the top of that fleet's distribution, and leaving aside PAK etc. If you believe such things you will find 4 to 6 times as many dead Allied AFVs as they actually had.

The average StuG might have killed 1.5 tanks, at the most generous 2 tanks, but not 8 tanks. If a unit claims 8 then they are either flat wrong or the unit is an unrepresentative outlier.

[Mr. Tittles]

'Had' meaning built and rebuilt? I still have not seen JasonC address the point of allied armor rebuilds and re-KO's.

Lets say that the soviets built 100K afv. They ended the war with 15K afv. But they HAD 150K knocked out. Its certainly plausible that the germans had some good ratios AND the soviets had some rebuilds.

[ May 05, 2004, 10:00 PM: Message edited by: Mr. Tittles ]

[CombinedArms]

Originally posted by Mr. Tittles:

But it is still an amazing stat. Even if they used up 10 stugs a month for 2 years, that works out to a 8:1 kill ratio. Even if that number were off 100%, any weapon system as cheap as the stug that could get such a positive effect is amazing.

I think that's the point...it's a little too amazing to believe. The Allies claimed an 8:1 kill ratio for the Hellcat TD. I also find THAT a little hard to believe. I've seen US TD battalions (36 TDs?) claiming 500 AFV kills and capture of 30,000 German soldiers. I find that hard to believe, too. Any unit that has fought well has a right to feel good about its performance, but it's dangerous to take their victory claims at face value.

I'm with JasonC in believing that while a few of the most extraordinary Tigers with the best and/or luckiest crews may indeed have posted 33-1 kill rates, this couldn't possibly have been the norm.

Normal sized US armored division had about 150 Shermans. The claim that 30 Tigers would kill 1000 tanks is the equivalent of saying that these 30 Tigers would in the course of their fighting lives single-handledly knock out the equivalent of 7 US armored divisions. Even if we allow for some of these Allied tanks being put repeatedly back into service, that just doesn't seem plausible. If the Germans had 100 Tigers involved in the Normandy campaign, as has been suggested above, that means that according to these claims they would have killed 3000 tanks total before those 100 Tigers eventually died. That's 20 armored divisions! But the allies didn't have 20 armored division in Normandy. And the Germans lost the campaign!

I think we have to think of claims like 1000 Allied tanks lost for every 30 Tigers as just inflated battlefield boasting--something not unique to Tiger crews...the Allies did it too.

[JasonC]

The Americans lost only about 1000 mediums in Normandy. They outscored the Germans there in absolute terms, when losses due to Falaise are included. The same is true in the Lorraine. The Bulge and Alsace probably were about even - at best 3:2 favor the Germans. They might have done somewhat better against the Brits, but not an order of magnitude better.

High German kill claims are fundamentally based on their performance in Russia, where they certainly did outscore the Russians overall. Russian tank losses for the whole war are around 100,000, dwarfing all other fronts. The portion of those coming in the period of the Russian offensive is around 3/4.

The Germans had only about 1 major weapon system for each of those kills, only half an AFV and half a PAK. So their AFVs certainly averaged above unity, in kills in Russia, and their kills were certainly higher than those achieved by the Russians.

But they are the same order of magnitude. The Germans produced half as many tanks as the Russians did, and as many tanks and PAK combined as the Russians had tanks. You can't have the average of an equal number of things each KO 8 to 33 of the opposite number, and still get the Germans to run out before the Russians do.

No Virginia, the Russians didn't lose 8 to 33 to 1 and make it up with overall odds. Their overall odds were no larger than 3:1. If their losses had run 8 to 33 to 1 they would have lost the war, decisively. They didn't.

To Tittles - the StuGs get damaged and returned to combat, the T-34s get damaged and returned to combat. Probably rather more of the former than of the latter (since the latter are more likely to be penetrated catastrophically and thus burn etc).

It does not make sense to compare vehicles on the StuG side, to instances into repair on the T-34 side. You can do apples to apples or oranges to oranges - vehicles to vehicles, or instances of going into repair on both sides. Vehicles on one side compared to instances of going into repair on the other is an apples to oranges comparison.

[ May 05, 2004, 10:02 PM: Message edited by: JasonC ]

[Mr. Tittles]

I dont think any afv system got anywhere near 33:1 just to set the record strait.

But unless someone is only counting 'brew-ups' or captured-damaged enemy vehicles, these probable over-kill stats emerge.

A uncontrollable fire that reaches either the fuel tank or the main ammunition area typically reduces a afv to scrap. Most other non-catastrophic damage can either be repaired or the vehicle can get a major sub-system swap or become a useful 'donor' for the mechanics.

In the east when the Germans were retreating, many kills could have been actually recycled. That is not to say that the Panzers and StuGs and others did not have tactical battlefield victorys. An example could be a section of two Tigers facing a company of 10 T34. The Tigers ambush them and hit and smoke up 4 vehicles (2 of which blow turrets). They hit three more which are abandoned and chase the rest away behind a hill.

The Tigers, seeing the need for ammo/fuel/etc retreat (they also have a minor injury from a glancing blow on the bow MG which breaks the radio operators shoulder).

They report 5 kills and 2 probables. The truth may be that only 2 were TWO, 1 long term and 4 short term. Seeing a tank smoke is not the same as a brew-up.

But a point is that the IMMEDIATE short term CM-like victory does follow the Tigers claim! So JasonC needs to see these claims and modeling AT CM's level!

Depending on the front/year/etc, I think Tiger Is could have MAINTAINED kill ratios of 3:1 to 6:1 depending on attacking/defending.

[ May 05, 2004, 10:48 PM: Message edited by: Mr. Tittles ]

[tecumseh]

Originally posted by JasonC:

Russian tank losses for the whole war are around 100,000, dwarfing all other fronts. The portion of those coming in the period of the Russian offensive is around 3/4.

and

Originally posted by JasonC:

Around a quarter of the Allied AFVs lost were Russian lights destroyed before midwar.

How many Soviet light AFVs were lost pre-Kursk to get this to work? Must have been a huge number.

[Andreas]

Originally posted by tecumseh:

How many Soviet light AFVs were lost pre-Kursk to get this to work? Must have been a huge number.

Almost the whole of the June 1941 Soviet tank fleet will count as light tanks (BT, T26). The exceptions would be about 2,000 T34, KV1, and T28. The whole fleet was about 23,000, IIRC. Almost all of these were lost. Add to these significant numbers of the T60 and T70 tanks produced (and many of them lost) pre-Kursk, and you end up with a huge number of Soviet light tanks lost that skew the figures tremendously.

[JasonC]

Tecumseh - the Russians started the war with around 20,000 light tanks. Around 12,000 T-26s and nearly 8,000 BT series. Most of these did not survive the opening summer, let alone the year, let alone 1942 as well. By Stalingrad they were essentially all gone. None were actually in production.

Instead they were making T-60s. They made over 6,000 of them by the end of 1942. T-70s came out a bit later, but around 5,000 had already been made by the end of 1942. The T-60 chassis was discontinued in 1943 and the T-70 production lines switched over to making SU-76s during the year, while making over 3,000 more T-70s.

Overall that comes to 34,000 light tanks, 31,000 of them either in the pre war fleet or produced by the end of 1942. There were plenty of T-70s alive at Kursk, but the T-60s were basically gone. If a few of those were left elsewhere (in cavalry unit e.g.), some of the 1942 T-70s had also been KOed. But leave the T-70s out of it (for a reason I will cover below). You get 26,000 T-26s, BTs, and T-60s fielded pre-war through 1942, basically all gone by Kursk.

One reason I separate these out is that these tanks were much easier to destroy than T-34s were. Significantly easier than even the T-70s. All the guns on the early war German fleet could kill them. The 37mm PAK the German panzerjaegers started the war with could kill them. The period when they died was also the period of German initiative, when they had the largest operational successes and racked up the largest casualty ratios in other categories.

The German weapons divide naturally between those that easily kill T-34s at range, and earlier systems that had trouble doing so. In the defensive period most of the weapons fielded could do this. In the offensive period most of them could not. The Germans transitioned their fleet to long 75s as a standard, and got rid of every smaller production type. The lighter chassis lines - Pz II, Pz 38, Pz III - went to making SP guns to meet this requirement. And German AFV production really only takes off after this transition occurs - a consequence of late mobilization of their economy.

The long 75 and better systems last until the end of the war. They do not become obsolete. The earlier systems do. They drop out of the killing side of the mix even if they don't drop out of the living mix, as the enemy fleet transitions to T-34s or better. The Germans fielded about 10,000 of the earlier AFVs. At the time of Kursk, only about a third of the fleet was still in this category. By the end of 1943, basically none of it was.

So, when looking at kill ratios for the later war German weapon systems - the long 75 StuGs, the Pz IV longs, the Tiger and Panthers, the later Jagds, the better 75mm and 88mm PAK and 88mm FLAK - the real force they faced and took their kills from is the Russian fleet after these early thin lights had already been KOed.

The light fleet was 26,000 to 34,000 tanks, with or without the T-70s, as transitional category time-wise. The Russians made 100,000 tanks during the war. Adding those they started with, they had 125,000 over the war as a whole. So basically a quarter to a fifth of the fleet - depending on T-70 accounting etc - was these early lights.

The Russians ended the war with a larger fleet than they had at the begining. The difference was about the same as they received in lend lease. So their domestic production and their losses were basically the same size. That means a quarter to a third of their losses were lights, concentrated in the early war period and mostly killed by the early part of the German weapon mix.

Leaving 67-75k losses to the late war German weapon mix. Thus, the relevant "universe" of available kills for things like Tigers etc, is this figure not the overall total. These kills then have to be portioned out among all the major German weapon systems of the later part of the war, AFV and PAK, plus the bulk of the infantry AT losses and AT mine losses, which mostly occurred during this period (as the Germans were on the defensive, and as much more effective infantry AT weapons were fielded late).

One can immediately see how implausible 45,000 Tiger kills and 68,000 StuG kills appear, once you notice this. Such figures are simply not possible. The Russians only lost 100,000 tanks overall, and they lost many of them before these weapons were even fielded, and they lost some of them to fausts and mines, and many of them to PAK.

And there is no reason to believe the StuGs outscored Pz IVs and all the later Pz Jgr models combined, or that Tigers outscored the much more numerous Panthers. Believing such numbers will quickly predict 5-10 times as many dead Russian AFVs as were available to be killed by late war German AFVs (after taking out some for fausts, PAK, etc).

The Germans made close to 40,000 of these late war AFVs, and most of them went to the east. 30k AFVs plus PAK etc at least as numerous can't kill only a portion of 67-75k AFVs and get average scores in the 8-33 range. In the 2 range is more like it, on average. That average is then internally composed of some higher than 2 - Tigers and Panthers - and some closer to 1 - Pz IVs and probably StuGs. Reasonable guesses might support 4-5 for the Tigers. But not 10-15, let alone 33.

[tecumseh]

I have read there were about 4,000 Soviet AFVs available for Bagration. Is this right?

It sounds to me like there should have been closer to 20,000+ available. Unless I underestimate the size of Bagration relative to the front.

Sorry for being off-topic

[Andreas]

Bagration was a large operation, but only covered 1/4 or so of the frontline. More importantly though, it was part of a concerted action that started in AG North sector in May (?), and then rippled south, with two tank armies being positioned in 1st Ukrainian Front sector against German AG North Ukraine for the L'vov-Sandomierz operation in July (1st and 3rd Guards), and one held behind 3rd Ukrainian Front Sector for the Iassy-Kishinev Operation in August (6th).

Additionally, independent Mech and Tank Corps, and Tank Brigades and Regiments would be distributed along the frontline where these major operations were to take place.

Finally, a lot of the available tanks were held in reserve, to make up losses and refill formations after their actions.

[tecumseh]

Thanks Andreas. I keep remembering that 4,000 figure like it was *total* AFVs, which is why I was thrown off by learning the Red Army starting with 25,000 +

I guess it was a small fraction of the total.

[K_Tiger]

@JasonC

Its totaly wrong to take the overall german Vehikel Produktion into your example. The overall losses due to Allied Arty, Air, Paks and Mines, was mutch higher as turned around, particularly after mid 43`. You can take out of your calculation all those german At-Assets who are never met an ennemy Tank face by face....this includes Partisan actions, abandoned Vehikels due to fuell shortages, mechanical breakdowns, overrunning...self destructing..ect. not to mention the losses due to Weather conditions in the cold/mud period.

The No.1 question is: How many German Tanks/any At-assets were responsibly for the kills of allied Tanks...of it I cannot read anything in your text, in addition, much text, no

force of expression.

Bring in a new calculation with actually combat ready tanks who possibly met and fougth against each other.... maybe than you will find a positiv result.

[Mr. Tittles]

I agree. This top down math is weak.

Also, I am trying to point out that at CMs level, immediate tactical battlefield victorys approach better numbers than long term accounting.

Other things such as 'is the vehicle defensive or offensive in nature?' plays in. The StuG, being a defensive weapon system, would have this in its kill ratio advantage. Any vehicle operating in an defensive role would have a better kill ratio than when it is in a offensive role.

[ May 06, 2004, 12:50 PM: Message edited by: Mr. Tittles ]

[Mr. Tittles]

A better analysis can be made with battalion level data over a long period of time. If you have Panzer Truppen 2, please turn to page 215 and 216.

The analysis is of 3rd Bn/Panzer-Reg 36 (14th PD)Jan-Oct 44. This is a mixed Bn of Panzer IV and StuG III.

Some quick highlights:

1. In these 10 months, the Bn had 61 days in action. A high of 17 in Jan and a low of 0 in Jul (rebuilding/rest?). They basically fought 1 out of 5 days on average.

2. TWO are 25 Panzer IV and 14 StuG and 1 HQ vehicle. This works out to 1.56 AFV per day in action.

3. KO'd /Destroyed Tanks/Assault Guns totals 175 and 34 respectively. These claims give approx 5:1 ratio. They also claim 161 ATG also.

[ May 06, 2004, 01:54 PM: Message edited by: Mr. Tittles ]

[Mr. Tittles]

1.44-2.44 XXXXVII 8. Armee Süd Tscherkassy

3.44 Reserve 8. Armee Mitte Tscherkassy

4.44 XXX 6. Armee Südukraine Pruth, Jassy

5.44 XXXX 6. Armee Südukraine Pruth, Jassy

6.44-7.44 (Refreshing) Reserve 8. Armee Südukraine Ceterini

8.44 Reserve 18. Armee Nord Kurland

9.44-11.44 X 18. Armee Nord Kurland

12.44-1.45 II 18. Armee Nord Kurland

2.45-4.45 Reserve 18. Armee Kurland Kurland

From January to June, 1944, the 14.Panzer took part in actions in the Kirovograd, Zveningerodka, Kishinevo, and Jassi regions. The 14th was pulled from the front in July, 1944 for rest and refit, and then sent to the nothern sector of the Eastern Front in August to take part in postional defensive actions in Kurland, mainly near Libau, until it surrendered with the collapse of Germany in May, 1945, when the Kurland Pocket fell.

[ May 06, 2004, 01:55 PM: Message edited by: Mr. Tittles ]

[JasonC]

And just as easily I can say -

Its totally wrong to take the overall Allied tank losses into your example. The overall losses due to German AT mines, infantry anti-tank, and PAK were much higher than the same turned around, particularly after mid-43. You can take out of your calculation all those Allied tanks that never met German AFVs face to face, because they fought the 75% of the German army that was infantry based, plus all those western ones lost at sea to U-boats, those KOed by the Luftwaffe in Russia, those that just broke down (and the Russians were notoriously bad about maintenance, and drove their offensives right up to logistical limits).

What you are really trying to say is that the average German AFV was destroyed or lost without killing anything, right? Well, in an average all the zeros go in the denominator. I know, how about if I only count Allied AFVs that killed AFVs on the other side? Or, shall I only consider the number of tanks KOed by the highest scoring single vehicles on the side? Where do I stop dropping out the mass of vehicles on the low kill side, just because?

Also, it is rather implausible that every tank that broke down or ran out of fuel also never fought an Allied tank. Wouldn't my denominator be rather low if I ascribed all the kills of tanks that fought for six months and later ran out of gas, to others that fought for 2 months and then were destroyed in combat? One might try to track vehicle-days rather than vehicles. But the answer one got would be about kills per vehicle day, not about kills per vehicle before it was lost.

Continuing to my parallel, I might easily say -

The No 1. question is - how many Allied tanks were actually killed by any German AFV, as opposed to those lost to all the other causes? Of that I can't read anything in your so eloquent post.

It is too easy to parallel the weakness of your own argument. If you think you have a better one, get your hands dirty with some numbers and do some analysis yourself.

Also, why should it count as a "positive" result, if one finds the KOs per Tiger are 14 and per StuG are 8, and not count as a "positive" result if one finds the KOs per Tiger are 4-5 and per StuG are 1.5-2? What is magical about the extra power of 3-5?

Both are evidence of successful weapon systems, since anything that on average KOs even one comparable enemy weapon is an above average weapon system. And no matter how high the number, the Germans still lost the war, largely to a power that outproduced them in tanks only by 2 to 1 - or at most 3 to 1 if the other fronts are included.

I repeat, there is no rational way to ascribe 8 to 33 to 1 kill ratios to average vehicles of a side that fielded a third to half as many AFVs - and as many major AT systems as the enemy had AFVs - and predict that the side so scoring will run out of the items exchanged at such ratios before the other side does. You can fiddle the denominator with assumptions, by a factor of two maybe, with a lot of handwaving and some apples to oranges and some one entry accounting. But not by a factor of 5 to 10.

There is in fact no sound reason to expect that German losses to things like air or fuel or breakdowns not induced by prior battle damage, will outweigh Russian losses to things like 20 million AT mines, 7 million effective infantry AT weapons, and 50,000 heavy PAK or FLAK. The latter might easily be larger, and is certainly going to be the same order of magnitude. So there is no sound reason to expect the AFV to AFV exchange ratio, to be higher than the overall loss ratio rather than lower.

And the overall loss ratio of late war AFVs vs. the Russians, is 2 to 2.5 times. Maybe if the German losses to extraneous causes are appreciable higher than the Russian, the AFV exchange ratio is 3 or 4 rather than 2 or 2.5. But it isn't going to be 10, and nobody has presented the slightest reason to believe it isn't just 2.

(Such a reason would have to numerically estimate both categories and prove the ratio of AFV to AFV losses was lower on the German side than on the Russian - and by a large factor, to matter at all for the conclusion. Nobody has presented any such analysis. Only hand waving and one entry accounting, seeing only one side of the ledger).

[JasonC]

As for what is "weak", propaganda claims are found to be false or misleading (true only of exceptional outliers e.g.) every time they are checked. Own side claims about the other guy's losses are found to be false every time they can be checked. By large factors, 2 or more.

Top down accounting is a standard principle in the practical world precisely because it cross checks mushy internal break down numbers. Nobody tries to close a ledger just by adding up item by item receipts, without looking at the bottom line and what is in the bank as well. Let alone by listening to the sales people around the water cooler and then multiplying whatever they say by the number of products or accounts or weeks in the period.

Any individual ratio in such an analysis can be off by a factor of 2. That is freely admitted. But all of them can't be. The end points are set in concrete (so and so many produced and lost, somebody ran out first). If one ratio expands another must contract. They check each other.

And the completely unreliable bottom up, watercooler anecdote methods regularly make predictions about overall outcomes that are demonstrably false. They count as theories, they make predictions about other aspects of the data, and those predictions can be checked, and they are wrong, every blessed time.

[Mr. Tittles]

Ill ignore JasonC's posts and continue the analysis.

So what about this supposed 5:1 ratio? Well, any fool can cast doubts about claims. No need for a speech about that. TWO is probably a very hard data point though.

The Bn claims 3.4 AFV destroyed per day in combat on average. Some days none and some days perhaps 6.6 or so at most. Not great numbers really.

But this battalion had a fluctuation in strength from damaged vehicles. It starts Jan with about 26 AFV. This gets whittled down to 1 runner on Jun 1. This 7.6:1 kill ratio (Taking data from JUST Jan to Jun) has cost the battalion its combat power from damaged vehicles.

The Bn is refit and from Jul to Aug only attains a 3.2:1 kill ratio. It again sees a decline in runners and combat power.

So whats it mean? In CM terms the Bn is probably kicking ass against T34/76, SU76, SU122 in the first part of the year but is recieving its share of gun hits, immobes, and spalling. Many vehicles are abandined but picked up later.

In the second part of teh year, they are about as half as effective. The T34/85, SU85, and other better AFV are taking the toll on the once mighty 75mmL48 vehicles.

[JasonC]

Let's look at Tittles example to see how poorly bottom up does compared to top down, or bottom up with full knowledge of the corrections needed, which only top down shows.

Suppose his case is typical and the claims are all real. Then 17,000 Pz IV longs and StuGs can be expected to kill 85,000 enemy AFVs. This leaves a negative number of AFVs to be killed by 6000 Panthers, 1350 Tigers, and 50,000 PAK. It is therefore false. The case is not typical and the claims are not real. The top down cross check immediately tells us this. Then we go look at the case and ask why?

The first thing we notice is that it is an apples to oranges comparison. It compares German claims of their field knockouts of Russians, with German reports of their own TWOs. And a claim of an enemy KO is not a report of an own side TWO. The two will in general differ, by some substantial factor.

The German staff officers during the war knew that own side claims were systematically high. They use own side claims anyway because they had no access to enemy records to get more reliable numbers. It was all they had to go on. But because they knew such claims were systematically high - from things like how long it actually took to stop a tank corps etc - they gave all own side estimates a 50% "haircut". In some cases after the fact we are able to check own side claims vs. enemy reports of losses. The 50% haircut is usually about right. Sometimes the claims are only 70% higher rather than fully double, but they are always high and substantially so.

Next consider that the German side of the ledger is friendly TWOs over a long period. That it is a long period helps deal with some repair issues. But some are left in repair at the endpoint, that are KOs in fact. In addition, some vehicles were KOed, recovered, repaired, and returned to service within the 10 month period. If the Russian side of the ledger we were comparing things to were TWOs, that would not matter - we'd compare apples to apples. But the Russian side is battlefield KOs, not TWOs.

If we want to know the ratio of battlefield KOs - how much more common it was for a Pz IV to knock out a T-34 than for a T-34 to knock out a Pz IV on CM scales - then we want not TWOs for the German side, but battlefield KOs. Which is going to be a higher number. Maybe 20% higher, maybe 50% higher, maybe double. We don't know which, we just know it will be higher, because not every battlefield KO turns into a TWO.

Either way of looking at it is going to reduce our 5 to 1 estimate substantially.

Next, is the sample representative? We hope it is because we hope it was chosen at random. It was selected from Panzertruppen because there was data on it. Is this likely to be a random sample? I submit it is not. Without ascribing any massaging or selection to the author, there will be "survivorship bias" in the sample. That is, we selected a unit that kept records successfully for 10 months and returned them, and that had a substantial number of running AFVs throughout the period, etc. Our potential sample could not include units cut off and overrun, piecemeal remnants, etc.

How important might such bias be? We don't know. It might be trivial and the sample basically representative. Or it might be higher than the average performance with bigger downside cases included, that our potential sample effectively excluded. Maybe by 1.2, maybe by 1.5, maybe by a factor of 2. We don't know.

From all of the above, the inference is that Tittle's case suggests an actual kill ratio for Pz IVs and StuGs somewhere in the range 1.5 to 3.0. The upper end of that range is more likely to be correct as to the performance of this individual unit, the lower end or middle may be more likely for the average Pz IV or StuG. The uncertainty left by the data is large. But there is no uncertainty about 5.0 being wrong.

Let's cross check again. 25,000 to 50,000 AFVs might be KOed by 17,000 StuGs and Pz IVs based on the bottom up estimate we arrived at. We can reject the high end as typical - it doesn't leave enough to be KOed by the top half of the German AFV mix. We can't reject the low end. The 3.0 figure might be true for the unit Tittles examined, but it can't be typical for all Pz IVs and StuGs.

The 1.5 figure could easily be true for the typical Pz IV or StuG. A middling 2-2.5 is possible, but would imply poor performance by German PAK and not much better than StuG-Pz IV performance by things like Panthers and Jagds. Which is possible, can't be rejected, but not particularly plausible.

How much did the 5.0 own side estimate tell us in the above cross checked analysis, and how much did the top down stuff tell us? I submit the bottom up tells us little. Certainly it is flat wrong if taken at face value. Corrected by a rough rule of thumb (the 50% haircut), it is no longer seriously misleading. We don't know how representative the result is; it becomes just one data point.

I hope this is illuminating.

[Mr. Tittles]

Hey, a speech was made anyway. Its sounds like JasonC is getting my relijun now.

Anyway, Lets suppose that there are 40 Panzer Battalions operating on the eastern front in 1944. They see 73 days of combat in the year. They get 3.4 enemy afvs per combat day.

Its a little less than 10,000 afvs for the year.

[ May 06, 2004, 04:25 PM: Message edited by: Mr. Tittles ]

[Michael Dorosh]

Originally posted by JasonC:

Suppose his case is typical and the claims are all real. Then 17,000 Pz IV longs and StuGs can be expected to kill 85,000 enemy AFVs. This leaves a negative number of AFVs to be killed by 6000 Panthers, 1350 Tigers, and 50,000 PAK.

Not counting panzerfaust claims, wildly inflated also in the Berlin fighting for just one example. Or Panzerschrecks, magnetic mines, grenade bundles, Stuka 87s with 3.7cm Cannon (Rudel alone claimed 512 kills), etc. etc...