Sorry for the title. I just couldn't resist.

TLDR - I created a google sheet that lets you see and compare average historical starting picks of players and groups of players - https://tinyurl.com/warzoneLadderData

There are a lot of things you can get from this data, detect some patterns from the top players, and maybe adjustments your early game picks. The data shows only pick 1 and 2, as you will always get at least one of them. I added information on how many games we got from each player so you can evaluate the significance of the data yourself. This is just the start of my analysis and with getting some feedback further analysis is very likely.

Example 1:
I try to play as best as I can, but only seeing this data I realized that I might unknowingly underestimate Greenland and Indonesia as an early pick, as I don't like slow starts. However, now I'm trying to reconsider those positions seeing that I'm way below the average of the top players


Example 2:
The higher your rating goes the more appreciation you have for East Us and Greenland, probably because it takes experience to realize that in fact, they could be a safe +5 bonus In turn 2. However, always remember that this is just an average. Even top players vary a lot, and everyone can have their own style.


Map:
With limits of google sheet, I also added some very basic geo visualisation:


Extra:
It's not included in the main sheet yet, but I'm also looking at the evaluation over time. Here we can see that with every game, a player started to play more certain locations e.g. Central America.


Edited 2/4/2021 13:33:13

thnx for this

Love this, a big ol +1

Really, really cool

It's really interesting to see the differences between skill levels. It seems that top players generally rely on less starting picks, as the picks with already a relatively high percentage are even higher for the top players.

Generally, West and Central russia are overestimated by lower rated player. Probably because they can have quite decent income from the start. However, it is also harder to defend because of the many double borders. Maybe top players are better in identifying these weaknesses and therefore choosing those less quickly.

Would you be able to do this for MTL as well?

Edited 2/4/2021 20:43:39

Incredible work!

The Greenland/Indo thing is actually unsurprising. In ps's little chat with Jz about Strat MME strategy from a while back (https://www.youtube.com/watch?v=tAUH3zwM13k), the sheer value of these picks came up early on. I strongly recommend watching that video if you're looking to improve on MME (and aren't already top-notch) because in that Jz explains the reasoning behind picking certain MME starts.

Edited 2/5/2021 17:28:31

Evaluation over time would be pretty cool for a player like ‘HotBeachBum’. He played many games from sucking to smashing it.

Evaluation over time for HotBeachBum:
I see that HotBeachBum focused only on picking bonuses +3


Edited 2/5/2021 08:48:22

The highest picking percentage for each territory (*for players that played at least 100 games)
The true king is The Young Wolf sticking with the Africa bonus however, with this information it might be easier to counter pick those moves.


Edited 2/7/2021 10:29:21

Lol the young wolf knows his triplepick xD

Gabriele knows what's up

Edited 2/5/2021 14:01:48

Gotta find yourself a girl as loyal as Alex is to scandiavia.

What about Young Wolf and west africa @plat?

Interesting project.

So if I'm understanding this correctly color-coding for each bonus is relative to other players. When I plug myself in I get a mild green for Scan at 38.9% but a dark green for greenland at 23% and west china at 21.2%. Meanwhile antarctica is a mild orange at 20%.

This would indicate I pick greenland and west china far more than the distribution of players, while picking scan more in raw games?

@Phobos: The colors are based on the value for you picking that bonus (as your first or second pick, when it's not wastelanded) relative to all the other values in the same row in columns C through O (you can check the conditional formatting rule by making a copy of the sheet; it's just a standard Color Scale rule), with yellow hardcoded by _ to be the average pick rate for all players (so if you're close to average, you'll be yellow).

So it's not strictly reliable since, e.g., the Blowfly column has just as much weight in that conditional formatting rule as the "All" column or each rating category. The players you pick in the right-side columns could, for example, make your West China preference look less exceptional if they're all players who, like you, go for West China. The shading itself also breaks down a bit since the color scale is based on max/min values, so if the distribution is low variance (say, the average player picks Bonus X 5% of the time, some pick it 6%, some pick it 4%) slight deviations will get highlighted with dark colors even though they're not really that notable. Thanks to the hardcoding by _, the yellow can be trusted but the dark reds and dark greens can get noisy.

But yeah, that's why your West China preference (21.2% when the average player picks it 7.6% of the time) is more exceptional than your Scandinavia preference (38.9% when the average player picks it 28.9% of the time).

Edited 2/6/2021 04:26:51

Nice work so far! It's pretty cool to see how picking patterns change with time and I'm curious if there are general patterns spanning most players at any given time (and if that changed over the course of the ladder). Also nice to see Blowfly making the chart for highest % picked bonus :P

Looking forward to see what else you have in store!

I added a new tab in the same sheet called counter picking. You can see % of getting certain starting positions and wining rate on them.

Example 1: in 67% of games of players <1500 they got their 1 2 3 picks and won 40% with it, at the same time top 1 players got 1 2 3 picks 40% of the time and won 67% of it. Top players try to counter pick their opponent as a strategy to get early information. I added an extra column so you can see everyone's personal score.


Edited 2/6/2021 18:12:01

I don't see 126 on the sheet? Other than that, great work!

yeah sorry already updated it with the 126.

Edited 2/7/2021 10:27:48

I added a new feature which is game search

You can use this search to see if there are any 2 identical maps but I think it's way more useful to find equivalent maps. You can set yourself relevant information e.g. certain starting positions, wastelands, or bonuses in which a wastelands are or in which aren't (as the exact location of wastelands doesn't always matter). The search requires at this point a manual input so let me show you an example:

Example 1 - easy
1. What do you want to find - was there ever a game where all territories of Antarctica were wastlands?
2. Get territories IDs - South Pole - 128, Siple - 93, Novolazarevskaya - 85, Scott - 20
3. Parse data by separating each value with the dot - 128.93.85.20
4. Insert the value to the column "Strict wastelands"
5. Result - https://www.warzone.com/MultiPlayer?GameID=19066680

Example 2 - find a similar game to the one you are playing
1. What do you want to find - I want to find a game similar to https://www.warzone.com/MultiPlayer?GameID=25622164
2. List all information that you find relevant:
- Bonuses with wastelands - Europ, Indonesia, East Africa, West US,
- Strict start - Sweden, Moscow, Siple
- Bonuses without wastelands (you don't need to list all just the one that you don't want to have them for 100%) - Central Amerika, Scandinavia
3. Get all the ids and format data
- Strickt start - 39.2.93
- Bonuses with wastelands - 4.16.17.9
- Bonuses without wastelands - 6.13
4. Insert the values in the columns
5. Result - 24445695, 16394551. 25352156, 25622164, 21453673

Remember that there might be 0 games matching your filter, that's why it's easier to have relaxed request focusing on bonuses instead of strict territories



Edited 2/12/2021 18:34:44

Incredible.