These were run on the same laptop that I used in the contest in 2010, which as I recall is slower than the contest virtual machines were. Bots which need the time may play weaker in my tests than they did in the contest. There are some.
Some of the opponent names here do not match the correct names on the official site. Sorry about that. These are the names I have them stashed under on my laptop.
300 games each against earlier versions on random maps. Not reliable for judging strength! Notice the very high draw rate because the bots play so much alike.
| bot | win rate | wins | losses | draws |
| oddshrimp2.3 | 58.7% | 103 | 51 | 146 |
| oddshrimp2.2 | 58.7% | 114 | 62 | 124 |
| oddshrimp2.1 | 61.2% | 174 | 107 | 19 |
200 games each against 10 opponents on random maps, 2000 games total. Unfortunately this selection of opponents doesn’t include many near oddshrimp2.4’s strength—it was better designed for judging earlier versions. I include average game lengths, which seem to tell more about style than about playing ability.
| bot | rank | win rate | wins | losses | draws | mean win | mean loss |
| GreenTea | 8 | 34.2% | 66 | 129 | 5 | 108 turns | 95 turns |
| dmj111 | 12 | 24.8% | 47 | 148 | 5 | 89 turns | 92 turns |
| medrimonia | 18 | 34.5% | 63 | 125 | 12 | 107 turns | 134 turns |
| rebelxt | 53 | 53.5% | 102 | 88 | 10 | 101 turns | 99 turns |
| fglider | 61 | 52.2% | 104 | 95 | 1 | 90 turns | 123 turns |
| eAshoka | 65-ish | 60.5% | 120 | 78 | 2 | 92 turns | 112 turns |
| FlagCapper | 91 | 64.8% | 128 | 69 | 3 | 97 turns | 153 turns |
| LudaBot | 122 | 69.5% | 137 | 59 | 4 | 72 turns | 168 turns |
| alocaly | 133 | 63.0% | 126 | 74 | 95 turns | 122 turns | |
| oddshrimp14 | 90-ish | 76.0% | 151 | 47 | 2 | 90 turns | 106 turns |
150 games each against 18 opponents on random maps, 2700 games total. You don’t need to run statistics or calculate ratings, you can pretty much read off oddshrimp2.4’s most likely ranking from this table. Don’t trust the answer too much, though, because the spread for each individual win rate is higher than people intuitively expect, and the set of opponents is too sparse to nail it down tight.
eAshoka was not an official contestant. Its rank was estimated by its author. I tested its strength myself, and I think the author’s estimate is good. Oddshrimp just happens to have its number.
| bot | rank | win rate | wins | losses | draws |
| iori | 2 | 25.3% | 34 | 108 | 8 |
| GreenTea | 8 | 35.7% | 50 | 93 | 7 |
| dmj111 | 12 | 18.3% | 27 | 122 | 1 |
| wagstaff | 17 | 36.0% | 53 | 95 | 2 |
| medrimonia | 18 | 39.0% | 55 | 88 | 7 |
| smloh | 19 | 65.0% | 94 | 49 | 7 |
| Neverstu | 28 | 41.7% | 62 | 87 | 1 |
| Manwe | 31 | 41.3% | 62 | 88 | |
| animatroid | 36 | 50.3% | 75 | 74 | 1 |
| mogron | 46 | 57.3% | 85 | 63 | 2 |
| deccan | 47 | 54.7% | 74 | 60 | 16 |
| rebelxt | 53 | 55.0% | 81 | 66 | 3 |
| fglider | 61 | 51.3% | 77 | 73 | |
| eAshoka | 65-ish | 65.7% | 97 | 50 | 3 |
| murrayr | 67 | 49.3% | 67 | 69 | 14 |
| Mistmanov | 77 | 49.0% | 69 | 72 | 9 |
| E323 | 78 | 54.0% | 79 | 67 | 4 |
| oddshrimp2.1 | 90-ish | 70.7% | 104 | 42 | 4 |
To get closer numbers for cases I wondered about. I kind of went overboard. Mistmanov and dmj111 score anomalously well against oddshrimp, but murrayr was a statistical outlier in the 18-opponent tournament. Oddshrimp in turn scores anomalously well against eAshoka and smloh. The anomalous scores have to do with style of play: One side makes mistakes that the other happens to be good at exploiting. Smloh’s big mistake, for example, is taking too many neutrals and leaving itself open—apparently that strategy works well against less aggressive opponents.
| opponent | rank | win rate | wins | losses | draws |
| dmj111 | 12 | 27.7% | 136 | 359 | 5 |
| smloh | 19 | 58.1% | 283 | 202 | 15 |
| Manwe | 31 | 41.4% | 204 | 290 | 6 |
| animatroid | 36 | 54.0% | 268 | 228 | 4 |
| eAshoka | 65-ish | 68.3% | 338 | 155 | 7 |
| murrayr | 67 | 56.6% | 261 | 195 | 44 |
| Mistmanov | 77 | 45.9% | 218 | 259 | 23 |
| E323 | 78 | 58% | 281 | 201 | 18 |
December 2013