WoT: Analysis of a losing streak

One of the “features” of being Aspergic is one tends to systemise things. One of the ways this manifests with me and World of Tanks is I record, on a spreadsheet, my daily record, both the overall record and what it is in each tank I play that particular day. This means I can look back and discover when my current losing streak began, and today I realised it is a month old. On 19th March I had a good day, a 10-3 record which had been preceded by a full week of winning daily records, in the month until that point there had only been three days played which did not have a winning record. On the 20th March however it appears things started to change, and days in which I had losing or drawn records became much more common. However, looking at the information more closely over this lengthy period andΒ a slightly different picture emerges. Look at the below totals (to US readers, beware the different date format) – warning much navel-gazing ahead πŸ˜€ :

20/3/13: 0-3
21/3/13: 1-3
22/3/13: 1-4
23/3/13: 8-9-1
24/3/13: 3-3
25/3/13: 0-0
26/3/13: 2-4

This is the first period of the losing streak, and I very much have a losing record in this period – by this point in time I was 15-26-1, and given this occurred over about a week it is not really a surprise that it has stuck in my memory.

27/3/13: 13-4
28/3/13: 4-2
29/3/13: 5-1-1
30/3/13: 10-8
01/4/13: 10-2
02/4/13: 0-0
03/4/13: 0-0

Things then turned around in the next week – in this week I had a 42-17-1 record, which is very healthy which ever way you look at it. Immediately thereafter, however, I have a hiatus (for good reasons), but in as much as that affected my gaming it clearly broke a good patch.

04/4/13: 4-4
05/4/13: 5-4
06/4/13: 6-3
07/4/13: 1-4
08/4/13: 3-4
09/4/13: 2-3
10/4/13: 8-8
11/4/13: 6-6

In the next week we have a struggle. My record is 35-36, almost a draw. I recall this week feeling very much like a grind, and it clearly was.

12/4/13: 8-17
13/4/13: 6-9
14/4/13: 0-0
15/4/13: 7-5
16/4/13: 7-4
17/4/13: 4-6

This is the 2nd Anniversary weekend, and subsequently, over which period I have a 32-41 record. A massive losing record, mostly concentrated in that first day of the special.

However, what is my record for the overall period? Well, Overall: 124-120-2, or a very slight winning record – but it has felt like I have been losing constantly for about a month.

How to reconcile this? Well in reality this is not very difficult, apart from the second portion of the period concerned I have had three weeks of losing records, which adds up. Moreover even though I do come in just about 50% for the entire period, I am used to winning 5-6% more than that – it seems a small difference but over several games it gets magnified. Finally in certain tanks I have had some quite long losing streaks, and once again that has magnified the effect of this difficult patch.

What to do about it? Well, that is very hard to say indeed. Part of the solution has to be to stop worrying about it – which I sometimes find difficult to do given how aware of the stats I often am. Beyond that is not to try to force something that is not there. In regular play my great weakness is still recklessness, and when I become a little impatient with how things are going I do try to do more. Also, in all honesty, I probably had a run of success prior to now that went on a bit too longΒ – something that likely caused be to be more reckless in any case.

I get the feeling that if this is an episode I could learn a lot from, to hopefully improve myself as a player. On the other hand I feel it could also sour the game a little for me. Hopefully not.

Update: Tonight I went 9-7.

Advertisements
6 comments
  1. seanas said:

    and lol, why do I have two names? Guess it’s a Friday morning fail ><

    • stnylan said:

      No worries – after I realised they were both from you I deleted the extra one πŸ™‚

      Very many thanks for your thoughtful comment, and you are of course entirely right. Nevertheless this quite extended month-long period going below my average is longest such dip I have ever experienced.

  2. seanas said:

    I don’t record my own results, but I do obsessively watch my daily stats via noobmeter, wot-news, vbaddict and the WoT mobile app, so I understand where you’re coming from πŸ™‚

    The simplest explanation is the random one: most of those sessions are just too small to give viable trend data. Given a 50% win chance, the probability of a 1-4 *or* 4-1 outcome over 5 matches is actually reasonably high: approximately 31% (if we assume the win chance is 50%, and it certainly appears to asymptote there for WoT, then for small samples it’s just a binomial problem). I know when I’m playing and on a big initial losing streak, I keep playing regardless (although not always on the same tank, often on my current favourite) because probability tells me that the streak WILL reverse itself.

    Likewise, if I find myself starting to get significantly ahead of the .50 win probability on any given day (after a sufficiently large number of games; for me that’s 20 or more) then I’ll consciously stop, because I know I’ll revert to the mean otherwise. It’s a bit more difficult when I get significantly behind the .50 win probability on any given day (after a sufficiently large number of games for me, again…) to keep going; usually I’ll stop then after a small winning streak – but I’m aware that that’s a non-rational response.

    We’re humans, we notice patterns, especially outlying patterns – and we have also a tendency to interpret random noise as a pattern when it isn’t. So we notice the big losing streaks – but the winning streak, those are skill man! Except that, well, they’re not – unless trend data indicates that we are in fact improving.

    The only time this is true is if something happens to bias the outcomes on a consistent basis – but that bias has to be evident in both the winning *and* the losing streaks. For me, the anniversary weekend had some bias: my current favourite tank is the Matilda, but the Matilda is horribly outclassed by derp-equipped tier V tanks – and as the anniversary gave a big incentive to play tier Vs, and thus derpy tier Vs, I found I had bad results on the Matilda. The solution, unfortunately, was to leave it in the garage the whole weekend. My Leopard suffered too, for the same reasons, but as I was trying to grind out the VK28, there wasn’t much choice but to play it.

    but yeah, TL;DR: binomial distribution + small samples = lots of random noise pretending to be patterns. I do feel you pain, however πŸ˜€

    edit: obviously a double comment fail >< this is what I meant to post.

  3. Gank said:

    I’ve been having a terrilbe week as well. We had a 5X weekend on the NA server and it always brings out the baddies. I use XVM stat and I can see the losses coming. Really, really bad players in the top tiers, or a really, really good platoon on the opposite team can really make a huge difference. It’s very discouraging especially when it extends for some time. As I’m very stubborn (read foolish) I usually keep playing whatever tank I’m losing in, desperate to get a win, and that ends up backfiring (6 losses in a row in my new Super Pershing, 45% win rate in my Pershing, etc).

    Argggg. Why can’t I just play for fun! Realistically after 9.5k battles no losing or winning streak of any kind will greatly affect my stats but I can’t help but obsess over them. Sometimes I think about starting afresh, but I don’t have the energy to grind all my tanks again and have spend waaaay too much on premium tanks for this account.

    • stnylan said:

      Playing for the fun is the end-goal we all aspire to. πŸ˜€

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: