Part II, The Wealth of Bullets
Scottish Economist Adam Smith wrote his epochal economic treatise "An Inquiry into the Nature and Causes and Wealth of Nations" Three Hundred and thirty three years ago, "a clearly written account of political economy at the dawn of the Industrial Revolution and" "widely considered to be the first modern work in the field of economics", but I will resist the temptation to state such grandiose a survey, and narrow it down to a subtitle such as:
An informal survey of localized ratings at ICC class C+/B- players, Fourteen Hundred Bullet at ICC, Part II:
I did some fairly hefty and detailed objective analysis of 50 other bullet players, plus another subsequent related 52 players, totaling 102 ICC players, and report the very interesting results here:
I set up a spread sheet and recorded in detail the results of all the twenty-five bullet chess players ranked proximate and just above me (1402-1406 elo), as well as the twenty-five players proximate and just below me (1402-1400). To sample this not only on one day, but across several days to randomize the data a little bit, I queried the set with the ICC command "rank dk-transform" and recorded it for that day, then in subsequent days, retained the records of those whose ratings were unchanged or were inactive, but then subsequently took careful note of new entrants, and thus in the days ahead took that snapshot several times to get fifty members of my data set. Not the stuff of pure science or allowing heuristical proclamations of averaging all the players in my class, but sufficient to take a very good snap shot of a 1400 bullet ICC chess players performance in other time frames in aggregate.
I not only of course recorded their bullet performance by wins, loses, draws, total games, best performance, but in careful detail their blitz, standard, 5-minute, and 1-minute performance as well--again, in full resolution.
From there I filtered the data via rank-sort so as to eliminate the 15-25% or so who had under-representative activity in each category (depending on what kind of chess it was), so as to get the best quality data set. Since most bullet players tend to also play blitz, this allowed me to establish an average measure of their blitz results among 43 players--or as much as 86% of them--for an average blitz rating of 1580 elo, standard chess play results among 25 players or 50% of them for an average standard rating of 1719 elo, 5 minute play results among 17 players or 34% averaging 1508 elo, and finally 1-minute play results among 19 players or 38% averaging 1058 elo.
Since some player tend to win one, or two, or three, or even play ten 5-minute, or standard, or 1-minute games and attain a perhaps uncharacteristically high posted rating for that category, they tend to thereafter completely discontinue that effort--so as to freeze for all to see their posted out of category result(s)--or alternately abandon the form with a poor result, dating from other years and thereafter remain inactive.
Nevertheless, not everyone does that, so I could establish not only average ratings, but detailed win to total games coefficients for some forty or fifty players, and so also determine the most competitive chess players in total. The idea being that a player 'A' who has won 470, lost 1093, drawn 65 among 1628 total games, and averaging 0.289 wins among all games is faced much stiffer competition than player 'B' who has W411, L269, D17 =697 at 0.590 wins, even if their bullet ratings are identical. Such a combatant is much more willing, by design, to loose many more games, and so is much more battle hardened and tested, in my humble opinion. More often than not player A carries this same tendency over to their blitz, bullet, and even standard games, and is a very interesting pattern to watch.
Along with that, as I was also saying, is that what happens more often than not, is that most instead concentrate all of their efforts upon bullet, blitz, or most likely of all, bullet and blitz in tandem, often resulting in the play of thousands or tens of thousands of games, deepening the solidity of my calculations.
My measures PER PLAYER averaged: 6,012 bullet games, 4,038 blitz, 153 standard, 1,246 5 minute, 6,330 1-minute, totaling an average of 12,873 games per player across the fifty players--of course, some were heavy users with 10,000+ or more blitz and bullet games, and some were new, with low numbers and were excluded to keep integrity of data.
* * *
Thereafter I knew that I had to take the assumed performance relative value of 1580 for blitz and 1719 for standard, to converge the data, and check for backward compatibility--of course, being all but certain to find some surprises in store, up or down!
So I then found the mean 1580 blitz player, using the same fixed window, whereby I then noted new members above and below that level, to sample the 31 blitz players (1580-1581 above, 1580-1579) and their respective bullet, standard, 5 minute, and 1-minute results. This time, I didn’t go through the cumbersome process of noting all wins and loses in detail, but only ratings by category, so as to speed up my query.
I found 15 or 48% bullet players averaging 1361 elo (eliminating the 16 who had insufficiently sizable games to get an accurate reading, et. al.), 16 or 52% standard averaging 1708 elo, 10 or 31% 5-minute at 1380 elo, and finally 7 or 23% 1-minute at 1004 elo.
And to round out my triumphate, of course, I of course also quickly sampled the 21 standard players proximate, above or below an assumed value of 1719 (1719-1722, and 1719-1718). Of course, while many bullet players play what are formally called 'rapid forms', not all standard players play rapids or even like to play any of them, but this is a good double check, since there are also some who play all forms:
I found 4 or 19% bullet averaging 1481 elo, 18 or 86% blitz averaging 1488 elo, 6 or 29% 5 minute averaging 1384 elo, and very lastly 2 or 10% 1-minute averaging 1004.
I know we could go on and on with this analysis, and shave it down, and talk about r-squared, and linear regression, but my main point or two is as follows:
I don't so much need to establish absolute truth, so much as to accurately and objectively survey my peers, and see where I need to be--more or less--in performance in other chess time frames and how I might formulate my goals going forward. I played 1300 bullet games, and am confident of the measure attained since as recently as November alone.
I found that 1402 elo bullet, as I suspected, does equate to a much stronger chess player than seemed proximate 1200 or 1300 blitz, and is often indeed associated with many 1600 and 1700 blitz players (this is an average, so of course does include some 1400 and 1500 blitz, in parity), as does 1580 blitz similarly equate with many 1800 and 1900 standard players (so of course some 1500 and 1600 standard). I found this in play, and now I have proven it. They are there.
With only 0.397% wins, I have had to fight many a ranking player, loosing a great many games 'almost won' but, alas, that is why 'they are' 1500 or 1600 bullet, they have a decisive edge, and can usually outfox you even under the burden or weight of their own sometimes not insignificant errors.
I found 1400 bullet much harder to attain than 1530 ICC blitz, and 1700 Yahoo blitz (2/12 or 3/8 as the case may be) and so will now play to get back to 1600 ICC blitz, where I started... but I am actually, I feel, truly stronger now than before so it remains a real open question as to how far I need to go there, to harvest similar partiy of chess skill and knowledge and effort.
Lastly, my honest pride at <40% wins at bullet yet having raised my rating about 22% in two or three months from a level, is as hard to do as I thought, not unprecedented, but difficult: only ten of those fifty had equal or better ratios (meaning lower), but alas, with my recent rate occuring much more like 33% instead of 40%, puts me not over 80% of my fellow bullet combatants, but more like 90%, a trend that I aim to continue.
My ICC blitz rate is now 0.333% in total, and is bettered by only 2 of those 25 members of the fifty, or above 92% of all combatants sampled.
As Dan Heiseman and Bruce Pandolfini both aptly say variously at chessCafe.com, 'try to play two players ranked above you, for every one below you'. I try to ONLY PLAY opponents above me, and at worst, only those ranked very near me if at all. Such a way is hard, but the wins are sweet.