whitherapathy
@whitherapathy
Followers
1,166
Following
30
Media
1,214
Statuses
3,279
Using Oryx data to infer as much as possible, but not too much. Also, on occasion, things to do with research, data and analysis. 🇬🇧 🇺🇦
United Kingdom
Joined June 2023
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Sant Rampal Ji
• 245191 Tweets
Corinthians
• 156405 Tweets
Raj Musix Kannada
• 154883 Tweets
Racing
• 129556 Tweets
Jets
• 108298 Tweets
Cazzu
• 103076 Tweets
Fate
• 97812 Tweets
#BAEMON_DRIP_OUTNOW
• 79263 Tweets
#WelcomeHomeJAYB
• 71441 Tweets
#GMMTVStarlympics2024
• 60940 Tweets
Nodal
• 60732 Tweets
PROUD OF YO JOURNEY
• 54630 Tweets
Happy New Month
• 51912 Tweets
Texans
• 42303 Tweets
Juanfer
• 41487 Tweets
Noroña
• 33224 Tweets
Alito
• 29433 Tweets
TAEMIN
• 27149 Tweets
Garrett Wilson
• 25163 Tweets
Garro
• 22104 Tweets
Hayırlı Cumalar
• 21678 Tweets
紅茶の日
• 19714 Tweets
LINGORM X VOGUE THAILAND
• 18419 Tweets
モンハン
• 16894 Tweets
キャラクリ
• 16869 Tweets
Botswana
• 16066 Tweets
#يوم_الجمعه
• 15522 Tweets
lil uzi
• 15095 Tweets
LOFFICIEL TH X LINGLING
• 14981 Tweets
Ramon Diaz
• 14155 Tweets
ヴェノム
• 13951 Tweets
PRABOWO Wujudkan NKRI Maju
• 13067 Tweets
Kompak Bekerja Majukan Negeri
• 12456 Tweets
Pinned Tweet
In a change to usual programming, I am going to try to explain the meaning of those grey bands in my charts (such as the one below, of
#Russia
's T-72 losses). It will be a long thread. I will try to make it as simple as possible. Apologies to non-native English speakers. 1/n
6
4
35
With 2023 nearly over, we can now count the total
#RussianArmy
losses recorded in Oryx this year. I have split the data into three main vehicle types (IFVs, SPGs, Tanks) and sorted by total losses (only 'captured' or 'destroyed' are included). 1/n
7
34
232
@wartranslated
Likely that most crew were alive on the way down. Which may have given them enough time to ask themselves why they voluntarily participated in an illegal war. Good.
7
3
221
This chart shows the composition of losses of Russian BMPs in Ukraine so far, based on Oryx data. There have clearly been some substantial changes over time. 1/n
4
10
155
By becoming so involved in the details of what is and is not permissible the US actually makes itself more complicit. Just give a blanket 'yes' and leave Ukraine to it, and everyone wins.
5
20
145
#Russia
is now pulling tanks deep from its reserves (AKA scraping the bottom of the barrel). The balance between tanks originating from the USSR rather than independent Russia is of interest. And here it is (based on Oryx data; tank origins as per the caption.) 1/n
1
11
139
There you go. It looks like direct conversion of T-80B into BVM, suggesting there might be no T-80BV left. If this is correct, in a couple of months the T-80BV might decline in loss data.
↑Также на кадрах визита Шойгу можно наблюдать, что танки Т-80БВ закончились – в Т-80БВМ начали перепиливать уже обычные Б-шки.
Учитывая массовую утилизацию в 90-е и огромные потери в войне танки просто кончаются.
10
31
187
4
16
123
I said I would come up with a model of how many deployed tanks Russia has, and how this might change over time. And here are the results. Details in the caption (please read before complaining!) but I will go through it all now. 1/n
9
16
121
Most of my posts simply report the facts about Russia's losses in Ukraine, as represented by Oryx data. But it's possible to do more than that. This is a 🧵of models predicting when Russia might run out of stuff. First: tanks. 1/n
4
16
110
This ternary diagram shows the three-way balance over time between losses of Russian BMPs, BTRs and MT-LBs in Ukraine. This balance has not shifted appreciably over time, which in turn suggests that Russia still has freedom of choice with its IFVs. 1/n
10
16
100
With even a minimum of effort Europe could help Ukraine produce 10 or even 20 of these per day. Imagine the consequences.
Ukrainian drones attacked Mozdok Airbase this morning, home to a number of Russian Tu-22M3 strategic bombers.
Seen here, a Ukrainian attack drone flies over North Ossetia, roughly 700km behind the front.
20
206
3K
4
8
93
I've implemented a change to my data which will show in the next round of updates to the charts.
@Rebel44CZ
appends the dates of loss to the image URLs in the Oryx data, and I will bow use these as the date unless absent. Taster chart of Russian tank losses below. No big changes.
2
4
89
Kursk and Pokrovsk has absolutely turned into a question of who blinks first.
Russian soldiers with no body armour or gear running and getting hit in the Kursk direction. Khorne group, who posted this footage, said Russians are sending all kinds of unequipped tramps to bolster the defences, indicating issues with reserves.
115
740
6K
5
8
94
I've updated the chart of 2023 losses of
#RussianArmy
IFVs, SPGs and tanks with today's data from Oryx (possibly the last 2023 update from them?) Note that the process of identifying dates of loss is complicated, imprecise and slightly biased towards more recent dates.
5
16
78
Sticking with SPGs for a bit, this chart shows the origin (Soviet, Russian) of lost
#Russian
vehicles over time. Oryx data, grouped in 4-week rolling windows. It oscillates between 75% and 100% Soviet stock. 1/n
2
8
67
Just the one chart left: an estimate of when
#Russia
's tank fleet might fall below viability (here defined as 500 tanks in service) based on loss rates derived from Oryx data, and some other assumptions. It's very unlikely that anything big will happen until late 2025. 1/2
9
7
65
With data up to 24 August, this is the composition of Russian tanks, by family, lost so far in Ukraine, based on Oryx data. The rise of the T-62 is now very obvious. 1/n
2
13
54
For
#Russia
's tanks, the future is inevitable. They will run out of stored hulls and, at some point, be left solely with new production. This ternary diagram shows the actual three-way balance of origins of tanks so far, and my projection of its future trajectory. 1/n
5
5
53
This plot shows the balance of
#Russian
losses across the three main land vehicle categories: IFVs, SPGs and tanks. it's based on Oryx data. The most recent data point is shown with a solid circle. Take home message: the data is not anywhere it has not been before. 1/n
4
5
53
First thing to note is that any model is only as good as its assumptions. And of course that... 2/n
1
1
51
Maybe it will be something like this? I suspect first increasing % BMP-2 as direct use of BMP-1 reduces and the old hulls are consumed in supporting BMP-2. And then inevitably, as old stocks run out, Russia can only field newly-built BMPs: the BMP-3. End.
4
2
48
But even at these rates
#Ukraine
needs more help from liberal democracies to win without continuing to suffer terrible losses, military and civilian. But win they will, as they can't be allowed to lose, and they are too brave and determined to do anything else. End.
0
2
46
I said I'd do a chart of composition of losses of Russian T-62 tanks in Ukraine so far. Having done so it's clear that it's perhaps a bit early, as loss numbers are low. 1/2
2
6
47
Is it a false impression on my part, or does Russia seem to be attacking with larger forces more frequently recently?
8
0
46
@hubertlepicki
@wartranslated
Yes, ideally you would not be able to see through a ship from the outside. My father was a seaman, and he taught me all the technicalities like that...
1
0
44
Remember the speech Hitler gave promising to multiply the weight of British bombs dropped on Berlin by 10x? Seem familiar?
Putin announced that in 2023, 140,000 drones of various types were delivered to Russian forces, with plans to increase that number tenfold in 2024. He also stated that the range of drone systems in Russia is expanding.
35
35
265
3
6
46
@Osinttechnical
I've watched this about 10 times now and still don't get bored. It's somehow symbolic of Russia's inevitable defeat.
1
0
43
Modelling a possible date at which
#Russia
's SPG fleet could become unviable is difficult due to the question of how barrel wear and replacement might play out. If we assume low barrel availability, high wear and that hull counts > barrel counts, we might get this. 1/n
9
8
43
Interesting thread on the T-90. Presumably new builds will deviate from the ideal spec more and more over time.
T-90M.
Yea its been a while since I said I will do that thread (like 4 months if I recall correctly :)))))). But well, the most modern Russian MBT, which is either extremely underestimated or extremely overestimated. How is it actually? Well, allow me to show you.
38
391
4K
0
5
44
I have shown my assumptions in the caption (that you just read, of course.) I use the recent RUSI
@Jack_Watling
estimate of 2060 deployed tanks at the start of the year. And my own estimate of ~4000 potentially salvageable hulls remaining at that time. 3/n
1
0
43
Following the recent update of Oryx data, we can start a new round of charts. First:
#Russia
's tank losses so far, by tank family and with data grouped in 4-week rolling windows. Very recently, and for the first time, seven (!) different families have been reported lost. 1/n
5
4
41
What is happening?! to
#Russia
's Infantry Fighting Vehicles (IFVs) in Ukraine? Answer: they are being destroyed, in varying proportions and at a huge rate. Here I have grouped into 11 families to make the data and chart slightly more manageable. It's still complicated though. 1/n
4
11
42
#RussianArmy
Infantry Fighting Vehicle losses by type, based on Oryx data. There are a lot of types, and a lot of losses. 1/n
2
5
40
I have a chart showing interest rate, inflation and USD-RUB rate over time. Will post at some point. Things like interest rates are much more important indicators than stuff like km2 of land lost.
The Bank of Russia raised the interest rate by 2%, from 16% to 18%, in an attempt to control inflation, which is much higher than the previous April forecast.
According to the regulator, “Inflation has accelerated and is significantly higher than the April forecast of the Bank
92
403
2K
5
3
41
It feels like we are at the peak of the second phase of Russian equipment use. 1st phase: moderate use of selected equipment, in anticipation of quick victory. 2nd: use literally anything in desperation. 3rd: constrained options, equipment types severely limited.
Remains of a Russian Kh-35 anti-ship missile landed in Novoshcherbinovskaya, Krasnodar Kraj in Russia.
13
81
1K
2
4
41
A direct appeal to
@AndreiBtvt
and
@Pion_2S7
(and any other Russian armour experts out there): to help produce a reliable summary of the capability of tanks lost in Ukraine, can you improve this scoring (1 to 10) of each variant?
16
3
40
With new Oryx data, up to 12 September, here is the composition of Russian tank losses in Ukraine so far. Data grouped in 4-week wide rolling windows, incrementing forward one week at a time. The squeeze put on T-72s is really clear. 1/n
2
2
39
Most notably, we are past the 'hump' of increased prevalence of BMP-1 that was apparent for much of 2023. BMP-1 is now below its expected range (based on the average of all the data, shown in grey) while BMP-3 is (just) above it. 2/n
1
0
40
@Jack_Watling
To work out the rate at which new tanks are added you need two rates: a genuine new-build rate (15 per month) and the rate at which old hulls are refurbished (60 per month). New build rate stays the same forever, but refurbishment stops when the stockpile runs out. 4/n
4
0
38
@Jack_Watling
To work out the rate at which tanks are lost, I use the average Oryx loss rate over the last 6 months, adjusted for a factor to reflect the % of all losses that Oryx actually records (which I assume to be 80%.) 5/n
1
0
38
@AndrewPerpetua
Only some of the vehicles that Russia is using are still in production. In the long run (or 'in the limit' if you want to be technical) it must be the case that the composition of losses will slowly show only those vehicles in production: tanks, only T-90s, BMPs only BMP-3. 9/n
1
0
37
@Jack_Watling
Russia is very much doing a Wile E. Coyote act at the moment, inflating its numbers with unsustainable levels of refurbished hulls. When the stockpile goes, that stops. 11/n
1
0
37
What about composition of losses? Charts like this one do not tell us how much stuff (in this case tanks) Russia has lost, but they do tell us how Russia is fighting, and they may show points of change (as
@AndrewPerpetua
emphasised the importance of.) 7/n
1
0
36
This ternary diagram shows the balance of
#Russian
losses recorded in Oryx data across the three main sorts in use: IFVs (in which I include APCs and IMVs), SPGs and tanks. 1/2
3
4
37
These are
#Russia
's losses of T-80s in Ukraine, based on Oryx data. Recent losses include only T-80BV/BVM and their Obr. 2022's. No sign yet of those T-80UD's that have been taken out of storage. 1/2
3
3
38
For example, if we know that Russia has access to 2k (or whatever) SPGs, and we can confirm that they have currently lost 1k (or whatever - figures are illustrative) we know that the fighting is in some sense 'halfway' to its conclusion. 5/n
1
1
35
These are the total recorded losses of various types of Russian military vehicles in Ukraine in 2024 so far, based on Oryx data. The SPG loss rate is, IMO, far in excess of any production/refurb rate Russia can sustain, no matter what stocks they have. 1/n
1
7
37
@Jack_Watling
Before I get bombarded with 'but that's too high/low': yes, I know, but you have to have an estimate. To incorporate the uncertainty, I allow these estimates to vary around the central estimate. For example, the new production rate may sometimes be 11 per month, or 20. or 16. 6/n
1
0
36
Although it describes what has already happened, Oryx data can also be used to predict what might happen in future. With a model specification as shown in the caption, this is when Russia might expect to run out of tanks. Middle of 2027? 1/n
5
2
36
To sum up. More data is better. All the great work people like
@HighMarsed
,
@waffentraeger
or
@Jonpy99
do looking at storage informs the picture immensely. It helps understand when Russia will run out of specific systems. 19/n
2
1
34
This is really interesting. I think it's this subtle, multidimensional attrition of Russian capability that is actually what will win the war for Ukraine. I'm as partial to a sunken ship as the next man, but in the end this is what matters the most.
@whitherapathy
@ArkanLopez
I analyzed the average range of SPGs in storage - it’s now 96% of pre-war range. (numbers based on
@HighMarsed
et al). The shorter range, the less effective an SPG, which requires more at the front, which requires more drawdown. Image below:
5
1
15
3
0
34
@Jack_Watling
For a model run, the central estimates are used to generate either a reasonable starting estimate (in the case of the initial stockpile and count of fielded tanks) or monthly estimates (for new builds, refurbs and losses.) 7/n
1
0
33
@Jack_Watling
Each model is different (in random ways): different estimates give different net change and so different counts of fielded tanks. But there's no point having one, arbitrary, model. So I do that 5,000 times. 8/n
1
0
33
That gives a sense of Ukrainian progress, which is good analytically speaking. But new production of vehicles means that, in a very real sense, there is no such thing as a halfway point. Numbers are dynamic. It's not a closed system. We can't only count Ukrainian progress. 6/n
1
1
32
I steadfastly maintain that a ternary diagram is the best way to look at three-way compositional data like this. So here it is. Currently (solid circle) the data is in a new happy zone. What will happen next is difficult to guess, but... 3/n
1
0
34
These two analyses tell us different things. But crucially neither tells us the thing that perhaps we're all most interested in: when will the fighting stop, because Russia can no longer fight. Both provide clues and indications of this though. 3/n
1
1
32
Broadly, analysis of loss data comes in two forms: analysis of actual counts and rates of losses (that is, how many vehicles have been lost and over what period) and analysis of composition of losses (that is, what % does each vehicle type contribute to the whole.) 2/n
1
0
32
@HighMarsed
Here the picture for Ukraine is more positive: some time towards the back end of next year, Russia may lack sufficient SPGs for them to... do the sorts of things that SPGs enable (I'm not an expert.) More likely that tactics will change than that fighting will stop. 5/n
1
3
32
Loss counts and rates can, if combined with reliable information on how many vehicles are available, indicate how far Ukraine has progressed on its necessary tasks of destruction. That's an important thing, but it's not the whole story. 4/n
1
0
32
@Jack_Watling
So that's it then. I am genuinely interested to hear alternative estimates for each of the model variables, with evidence for why they are better.
@whitherapathy
runs an on-demand modelling service for those with sensible suggestions 😀. End.
4
0
32
@Jack_Watling
I then take the central estimate from those models, and also the ranges of values produced across all 5,000 of them. This is what creates the pink bands. They are ranges within which 99%, 95% and 50% of the model outcomes lie. You knew that ofc: you have read the caption 😉 9/n
1
0
32
While Oryx data doesn't capture every one of
#Russia
's vehicle losses (personally, I think it's ~80% of them) we can still infer useful things from the rate at which losses are recorded there. These are losses of tanks in 4-week rolling periods (÷ by 28 to get daily rates). 1/n
4
6
31
@Jack_Watling
If the number of refurbs in the model is too low, decline is closer, but Russia has more tanks in the short term. It, like everything in the real world, is a trade off. 13/n
1
0
31
@Jack_Watling
Anticipating complaints from those who want me to say that Russia will run out of tanks tomorrow, note that there is a trade off. If the number of refurbs per month is too high, that stockpile will last longer and the decline comes later. 12/n
1
0
31
As per
@highmarsed
Russia probably had about 7k tanks potentially available (active and recoverable in storage) before they started. We can use Oryx-derived loss rates and a simple linear model to predict when they get below viability (500 tanks.) 2/n
17/ Conclusions:
I: Russia had at least 7000-7500 full tank hulls at the storage bases and BTRZs in 2021.
II: There is physical space for 3000 tanks in garages, but the number is likely much lower (<2000), since some garages had IFVs, ARVs, trucks and other vehicles.
3
20
461
1
1
31
A couple of more niche charts. First,
#Russian
artillery losses by barrel family (h/t
@verekerrichard1
for the classification of families). Relevant to the debate about whether barrels or shells are the limiting factor in artillery perhaps? 1/n
1
5
30
In the absence of that, I think we need to look mostly at the composition data, to see how far along the path to losses being of new production only we are. It's pretty clear that we are a long way off that point. I think this may not be a welcome fact, but it is a fact. 18/n
1
1
29
@HighMarsed
Looking at artillery more broadly, the question is what is the limiting factor: shells or barrels? Based on a very simple estimate of the potential number of rounds that Russia can fire, it looks like it's the rounds that matter most. 6/n
2
1
30
What if Russia is not exaggerating the rate at which it builds new tanks and refurbishes old ones? The claim is ~150 per month. According to
@RUSI_org
this is ~20% new build. Based on that, this is what the count of fielded tanks might look like over time. 1/n
3
2
31
For looking at the (very basic) model of potential fade-to-zero of
#Russian
SPGs some of you may prefer this version. It shows the quantiles of the points, rather than the regression lines themselves. 1/n
5
3
30
@Jack_Watling
the only way that is not true is if the refurb rate is low, AND the stockpile is smaller than I assume here. To my eyes, there does not appear to be any way in which Russia runs out of tanks for years to come. Their force may become unviable though. 14/n
2
0
30
The quicker that Russia gets to its maximum production level, the higher the number of vehicles it can field in its equilibrium state. In that state, counting losses stops telling you much about Ukrainian progress (though changes will shift the equilibrium.) 14/n
2
0
29
At the other end of the charts there are some unusual types, many with just 1 loss. These include the recently released 2S34 Khosta. There are more Terminator movies than there are destroyed BMPT 'Terminators'. 3/n
2
1
30
Same, but for SPGs. This suggests the middle of next year. Sounds about right. End.
2
1
30
Overall pop pickers, the most common
#Russian
loss was the T-72, just ahead of the BMP-2 (I have checked, and Alan Freeman is among the few 60's and 70's BBC radio DJ's not plausibly accused of sexual assaults, so I can post his image here.) 2/n
1
1
29
@AndrewPerpetua
They show strain on Russia's military. They also, as I have pointed out, show progress also, on the path to inevitable end points. What do I mean by that? 8/n
1
0
28
@AndrewPerpetua
The issue we have is the new production. The logic is that the number of vehicles of a specified type that Russia can field will reach a minimum when, and only when, new production is equal to the rate of loss. This is the 'one in, one out' position. 11/n
1
1
28
With that in mind, the importance of good estimates of Russian production is very obvious. it determines much about their fighting capability. And of course as newer stuff is usually better, the higher the % of deployed vehicles 'in production' the more capable the army is. 15/n
1
0
28
@AndrewPerpetua
If we track changes in composition we can also see how close we are to this logical end point. That too is nice, but it too is limited as an approach, because if the composition changes but the rate of loss is low, it doesn't matter much militarily. 10/n
1
0
28
@AndrewPerpetua
But this could happen at any time. For tanks, it might happen today, or in a year, or it may have happened some time ago. It's the balance between new production and losses that determines when the size of the Russian fleet will stop changing. 12/n
1
1
28
Genuinely don't know how
@Rebel44CZ
would report it if Ukraine captured a military museum in Russia. Are they exhibits, or are they war materiel?
4
0
29
Updating the code used for my charts. Had the honour of adding a line to record T-54's. Should I do the T-34 line while I'm here?
4
0
28
When a proper, full history of this war if written, the chapter on the Krynky fighting will be fascinating. Why can't or won't Russia gather together enough force for a one-off operation to end it? Why decide to allow your forces to be bled so successfully? Absolutely bizarre.
Battle of
#Krynky
equipment loss numbers as of 03 May 2024.
In summary:
57 Ukrainian 🇺🇦 losses vs. 244 Russian 🇷🇺 losses
Spreadsheet showing the losses in detail:
1
13
108
0
1
27
@AndrewPerpetua
And, importantly, it might happen at any number: 100, 1000, 5000 tanks. We don't know, and we can't count down to an end point of zero capability, only to the point of fixed capability. All we know is that the balance of losses must shift to show only in-production losses. 13/n
1
0
26
BMPs are the largest family of Russian IFVs. This plot shows the composition of BMP losses over time, in 4-week wide windows of data incrementing forward by one week. I don't look at variants of BMPs, but
@verekerrichard1
does if you want that detail. 1/n
1
3
28
When might we expect
#Russia
to run out of tanks? A simple model (explained in the caption to this plot) suggests that it probably won't be before autumn 2027 if current loss rates continue. There's about a 1 in 4 chance of it happening before the start of 2027. 1/n
3
3
28