Tuesday, April 7, 2020

To study openings part 4

I concluded my previous article with the "funny" note that I became a chess-celebrity in Belgium thanks to this blog. There are a number of stronger players in Belgium but it is very hard to become famous by just playing chess. However maybe even greater than this fame is my reputation of being a theoretician which even goes way further back in time than this blog. I can't formulate it better than the already earlier referred report of the KOSK: "Brabo... is feared as chess-player with his good/great/ambitious/exaggerated opening-knowledge". Besides a couple of days ago I suddenly got an email from a Dutch expert asking me to help him with the Graf-line of the Spanish opening so it seems my reputation doesn't stop at the boarders just like the corona-virus.

I hear this song for sure already 2 decades. I always considered it nonsense but I did like it when some of my opponents tried to play the most ridiculous openings just to avoid my "encyclopedic" opening-knowledge (remember e.g. my article scholar's mate). Yes indeed this isn't a good example as I even knew some "theory" about the scholar's mate. Therefore I thought it would be a good idea once to check how much I really know about openings. Who exaggerates or maybe the truth is somewhere more nuanced?

Last week I wrote in a comment that I was busy with a large research and I didn't lie as I spent many hours to investigate the openings of all the standard games I've played in the last 20 years. In total there were 722 of them so it took a while to define from each of them if I had some foreknowledge of the opening. I was lacking time so I had to spread the work over several days  but I believe the final result made the effort worth. However before I start showing the results, I first need to explain to the readers what kind of definition I used for an opening.

There exists no consensus between players about what an opening exactly is. Of course you can consult Wikipedia or Eco but that is not useful for any statistical research. So I used in the end the same arbitrary rule as I did in my article part 2. An unique opening starts from an original position of which maximum 100 master-games (one of both players has a +2300 rating) exist in the megadatabase. So for each of my 722 games I checked at which move the opening started based on an opening-book created from all the master-games existing in the megadatabase. A screenshot explains it always better than 1000 words.

This position occurred on my board in my most recent match in the Belgian interclubs against the Belgian IM Stefan Beukema. Based on the 100 master-games-rule I consider 8.Na4 and 8.Ne4 as 2 different openings. I think everybody agrees both openings should be treated separately.
Next I divided the openings into 3 categories based on my own opening-knowledge:

Category 1: I never played this opening neither did I check it with an engine. I never read about it in a book or forum. Maybe I encountered it in a few online blitz-games but I didn't pay attention to it and just clicked to the next game. Stefan chose 8.Na4 in our game which fulfills all those requirements. I responded with the weak 8....Nfd7.

Category 2: I know something about the opening but it is very limited. I checked maybe some lines during a preparation of a game. Maybe I read something about it in a book or forum. Possibly I played already a standard game with this opening but it was a different line and I didn't check yet the other lines. I have found no games of Stefan in the megadatabase with 8.Ne4 but 8.Ne4 was still part of my preparation. So if he had played 8.Ne4 instead of 8.Na4 then I would've considered the opening as category 2 instead of 1.

Category 3: I've made once an extensive analysis of the different lines with an engine. It can be that the opening popped up in one of my earlier games like I explained in part 2 but it could also be that I once played a correspondence game with the opening. So this doesn't mean necessarily that I am fully up to date of the theory to categorize the opening as 3. Below screenshot is a nice example of such scenario. In the last round of Cappelle La Grande played a month ago, I got 9...d6 on the board. I never encountered it before in any standard game but my opponent had bad luck. I was able to reproduce my analysis made in 2014 till move 18 which I published already on my blog see my game against Roel Goossens in my article fashion. As I introduced a strong novelty at move 15, I think we can consider this a category 3 as opening despite I haven't made any recent updates to the analysis in the last 5 years.

So we have an understanding about how to differentiate between openings and categories. It is time to look at the numbers. I invite people not trusting my data to write me a mail so we can check together and discuss. I do admit that there were a few tricky cases in which you could argue about the interpretation of the category/ opening. Anyway we are dealing here with input from more than 700 games so I think we can safely state that those few boarder cases won't make any serious impact on our statistics. The first table shows per year clearly the number of games played per each of the 3 categories .
The average for the last 20 years was just below category 2: 1,8. The split between the categories is 38% for category 1, 43% for category 2 and only 19% for category 3.

My first year 2001 is clearly the worst which is no surprise as I was rather inexperienced at that time. However I was a bit disappointed to see that my opening-knowledge almost didn't expand in the 2 decades. I've spent an enormous amount of time at analyzing openings. Why is that?

Well the first reason must be that the theory exploded. I already remarked this in 2013 see revolution in the millennium and the speed of it only increased in the last couple of years. Some openings split into 2, 3 or even more openings. Players like to follow fashionable openings so one is always obliged to learn new stuff. However when looking at the figures it is also important to check the mix of the opponents. Therefore I made a new table but this time linked per elo-category. This shows a very different picture of my opening-knowledge.
We notice that the number of lower-rated players have drastically increased over the years while the opposite happened for the higher-rated players. No I didn't get scared but it is the logical consequence of playing only in the neighborhood after the birth in 2007/2009 of my children. That is why I chose for smaller tournaments close to home like the club-championship which has not very strong opposition (see inactivity).

We also notice that my opening-knowledge has little or no influence below 1800 elo. In part 1 I already made the remark that 1800 players know very little theory of openings and this new study just confirmed it.

Next I also find it remarkable that we see a clear increase of opening-knowledge till 2200 but higher it again decreases. No I don't play junk openings against +2200 rated players nor do +2200 rated players ignore theory of the openings. At the higher echelons we see players start to vary their play and try to get their opponents out of book. So it is not like those strong players know less theory at contrary as it is increasingly studied.

Finally the last interesting aspect of above table is that we can definitely see an improvement of my opening-knowledge over the years. I am glad to see my efforts payed off. An increase of 1,7 to 2,3 in the rating-group 2200-2400 seems not much but very few know how much work you need to achieve such gain.

I still have one last question. In which extend did my reputation harm my results? I mean can we see a difference of the average category between games played against people knowing my reputation and games played against strangers. Before we have a look at this, we first need to check about how many games we are exactly talking. Below table shows per year how many of my opponents are aware about my reputation or not. I again agree there are some cases which I wasn't sure about but as there weren't many this will not change the final view.
The difference between the first and the last years is evident. The year 2006 is special as in that year I played the closed master-tournament in Bruges against players knowing me very well. That influenced the average of the year a lot. I grew up in the same province which explains why they were much earlier than others to know my strengths and weaknesses.

In 2012 I created this blog and this must be the reason why suddenly I started to meet much more often opponents being very well aware about my habits. In 40% of my most recent games I experienced that my opponents consulted this blog and didn't hesitate to use my articles against myself. It also explains why I was eager to play again abroad (Netherlands and France) without having this disadvantage.

On the other hand I see my opponents use often very different strategies to cope with the bonus-information available on this blog. Sometimes a cure is worse than the disease as you never know exactly what I have studied. Preparations start to play a big role above +2200 elo so I think it is therefore interesting to focus on the 2 highest categories to see to which effect foreknowledge about myself influences the openings. We know from the previous table that the group of people knowing me well, becomes from 2014 onward substantial so I will only look at the years 2014-2020.
In other words I lose about 0,3 averagely in both categories when opponents know me well compared with strangers. It is not pleasant but I think too many readers would find it a pity that I stop because of it with blogging. Besides by playing more abroad, I can maybe avoid this disadvantage more regularly.

I want to conclude this special article. It is clear that I do know quite a bit about openings. I won't deny that this is much more than the average amateur. However I still encounter too many times openings which are totally new for me so I am still far away from the level of professional players or even Belgian international masters like Steven Geirnaert, Stefan Docx, Bruno Laurent... Now I don't think this article will change much about my reputation as it is almost impossible to get such thoughts extinguished.


No comments:

Post a Comment