Round 8 Pairings at FKB US Ch

Running Chess Tournaments
1

So as everyone knows, it’s a nightmare to do eighth round pairings in a small tournament, and there is virtually no way to make everybody happy. Here’s an issue that came up at the Frank K. Berry U.S. Championship.

Please see the Monroi web site for the wallchart.

The difficulty arose when pairing the 4.5 score group. Obviously one of them has to play up against Shabalov, but the problem is that five of the seven are due White, and Gulko and Ehlvest have already had four Blacks in seven rounds. Our issue was should we give Ehlvest a fourth Black in five rounds (something we absolutely hated to do, but ultimately decided to do) or give Gulko BWBWBWBB and make a much worse odd man choice (Becerra instead of Gulko). There was no good way to pair the score group (or indeed the bottom groups), but what would other people have done?

Alex Relyea

2

Are you using USCF or FIDE pairing rules?

Tim

3

USCF.

Alex

4

Although it is tempting to put the results into a pairing program, I’ll take a shot at embarrassing myself publicly and give my non-computer-assisted pairings and logic.

Since the only two 5.5s have already played, the initial pairings would be (ignoring color)
2 (5.5) - 12 (5)
6 (5.5) - 3 (4.5)
4 (4.5) - 11 (4.5)
7 (4.5) - 13 (4.5)
9 (4.5) - 14 (4.5)

3 and 4 have already played 2 and 6, so the next in line for 6 (due black) is either 7 (2667 due black) or 9 (2660 due white). Since there is an overabundance of 4.5s due white, 9 should be pulled up instead of 7 (a 7-point transposition), resulting in (for the top two boards):
12-2 (both due black, 2 is higher ranked)
9-3 (colors work)
and ignoring colors for the rest (7 and 11 are due black, 13 and 14 must have white, 3 and 4 have matched colors for rounds 5-7, with 3 having white in round 4 and 4 having black).
3-11 (colors work)
4-13
7-14 (colors work)

However, 3 has played 14, 4 has played 13 and 7 has played 11.
Since colors work for 7 and 14, and that is the natural pairing, you could leave them alone. To avoid having 4 play 13 again, you would then do a switch, resulting in:
13-4 (13 must have white)
4-11 (colors work)
14-7 (colors work)

This makes a 3-point transposition or 3 and 4 rather than the actually used 3-point transposition of 13 and 14 (which left alone the 3-11 pairing where colors worked). Either one is initially mathematically equal. One reason for doing the 3-4 switch is based on considering 4 slightly more due white than 3 is. Also, the other half of the switches are 12 points with an 11-13 switch, and 16 points with a 4-7 switch.

In 29E.4 - #1, it talks about due colors when pairing two players, but interestingly enough it does not say the two players are actually paired against each other. That alone can be used to justify giving 4 white instead of 3. I think that would be stretching the rule though since the intent seems to be geared only towards players paired with each other. Although it would be nice to give 4 a white rather than 3, it seems a bit questionable to use that as the reason for doing the 3-4 switch rather than the 13-14 switch.

The rulebook ranks transpositions’ desirability when the transposition differences are different, but doesn’t cover them when they are the same.

All things considered, since the 3-4 and 13-14 transpositions are both 3 points, and the 11-13 transposition (the other half of the 3-4) is 12 points versus the 4-7 transposition (the other half of the 13-14) being 16 points, mathematically the 3-4 transposition seems marginally better.

That would result in boards 3 to 5 being
Perelshteyn-Kaidanov
Ehlvest-Ivanov
Becerra-Shulman

Since the actual pairings also follow a top-down logic, and they are so close, I would say that if the tournament has been using top-down logic already (which I didn’t check) then the actual pairings would be consistent and appropriate.

5

Just to clarify, though Onischuk received his due color because he was higher ranked, that consideration only came into play because he had exactly the same color history as Kudrin. Otherwise 29E4 #1 would have governed who received which colors.

6

I didn’t find this as difficult as I thought I might.

Standings after 7 rounds, in order by score, then rating:

                              color      due     has already been
                             history    color     paired against

 1.  2712  Onischuk    5.5  WBWB WWB      B          2,4,5,8,9
 2.  2671  Shabalov    5.5  WBWB WBW      B          1,4,5

 3.  2612  Kudrin      5.0  WBWB WWB      B          4

 4.  2686  Kaidanov    4.5  BWBW BWB      W          1,2,3,10
 5.  2683  Ehlvest     4.5  WBWB BWB      W          1,2,9
 6.  2667  Shulman     4.5  BWWB WWB      B          8
 7.  2660  Gulko       4.5  BWBW BWB      W          10
 8.  2623  Ivanov      4.5  BWBW BWW      B          1,6
 9.  2611  Perelshteyn 4.5  BWBW WBB      W          1,5
10.  2608  Becerra     4.5  WBWB WBB      W          4,7

Once Onischuk has been paired against Kudrin, the remaining eight players are mostly due white (5 due white, 3 due black).

So, among the remaining eight, we certainly wouldn’t want to pair two due-blacks against each other, as this would further worsen the scarcity of due-blacks.

This means that Shabalov, rather than being paired against the top 4.5, should be paired against the top 4.5 who is due white. That player is Gulko. (Shabalov has already faced both Kaidanov and Ehlvest.)

Side note: If, in pairing a 5.5 against a 5.0, Shabalov (rather than Onischuk) were paired against Kudrin, the exact same thing would happen, since the colors are the same either way, and since Onischuk (like Shabalov) has already faced both Kaidanov and Ehlvest.

So far we have:

A. Onischuk vs Kudrin (both due black, Onischuk gets black)
B. Shabalov vs Gulko (Shabalov gets black)

If the remaining six are paired strictly top-half vs bottom-half, we have:

C. Kaidanov vs Ivanov (Kaidanov gets white)
D. Ehlvest vs Perelshteyn (both due white)
E. Shulman vs Becerra (Shulman gets black)

Note: All pairings are shown here with the higher-ranked (higher-scoring or higher-rated) player on the left.

But Ehlvest has already played Perelshteyn. Do we switch Perelshteyn with Ivanov, or with Becerra? Either way, it’s a 3-point rating switch.

Switching Perelshteyn with Becerra would result in an Ehlvest vs Becerra pairing. This is not good, as both Ehlvest and Becerra are strongly due white. (Becerra has just had 2 blacks, Ehlvest has just had 3 out of 4 blacks).

So the other transposition is better, and we end up with:

C. Kaidanov vs Perelshteyn (both due white, Perelshteyn gets white)
D. Ehlvest vs Ivanov (Ehlvest gets white)
E. Shulman vs Becerra (Shulman gets black)

Hmm, upon reading (rather than just skimming) jwiewel’s post, I see that we have come up with the exact same pairings, using the exact same logic – except that I feel more strongly than Jeff does that an Ehlvest-Becerra pairing should be avoided because of very bad colors.

Bill Smythe

7

The kind of analysis done here by Bill and Jeff is exactly the kind of thing we look for in our ANTD and NTD essay exams. They combined attention to detail with and understanding of the nature of pairing rules, above and beyond the wording of the pairing rules.

Tim

8

Well, the upshot of everything is that we had a complaint from a player about the pairings. I contacted a special referee (BTW, are the special referees listed somewhere on the new website? It might be handy.) and he approved the pairings the way the computer suggested. If we go with the pairings that both Bill and Jeff suggested, we transfer the problem of having five Blacks in eight rounds from Ehlvest to Kaidanov. Obviously there is no good solution, but the special referee said that he knew of no rule that kept a player from getting four Blacks in five rounds. I agree that this was a horrible solution, but I don’t know that any solution would have been more fair.

Alex Relyea

9

That doesn’t make the pairings wrong. Any GM assigned 5 out of 8 blacks is likely to complain, but in this case it was unavoidable. It’s just a question of which GM is so paired.

Did he say the computer pairings were best, or only that they were legal? Many special referees (and many chief TDs, for that matter, when considering changing the pairings made by their computers) will simply go along with the computer, unless something seems grossly wrong.

By the way, I agree that the pairings you used were legal. And so does Jeff, I’m sure.

True, but at least Kaidanov would still have had only 3 blacks in the most recent 5 rounds, not 4 out of 5 as Ehlvest did.

I’m sure there’s no such rule, and it’s not as bad as assigning 3 blacks in a row, but when I see BBWBB I cringe.

The pairings suggested by Jeff and me wouldn’t have been better and more fair?

Still, I have to keep reminding myself that the pairings might have been made under pressure, without a lot of time to consider all the possibilities, and that the special referee might have received only an incomplete account, perhaps by telephone, and perhaps under time pressure.

I also should remember that I myself have often suggested (on these forums) that TDs should be wary of changing computer pairings. Computers often see things human TDs don’t, especially if a lot of looking ahead is involved.

Bill Smythe

10

One thing Tim Just has often said is that if a number of NTDs look at pairings then it is quite possible that every one of them will come up with a different pairing. Also, every one of the pairings will be legal and logical. As I said, the ones you used were consistent with top-down pairing logic, and that is just one of the reasons that make them quite justifiable.

By the way, Bill and I do have a disagreement.

I felt that giving somebody four blacks in five rounds is “not good”, but did not see that alone as being a sufficient reason to change the pairings (either player with black would have five blacks in eight rounds). My change came due to the lower point difference in the other half of the two switches (4-7, 11-13), and I only looked at that because the two switches (3-4 & 13-14) were tied in their point difference and there was nothing in the rulebook that I found that specifically addressed which pairing is better if two had the same point-range for the transposition.

Bill was more bothered by giving four blacks in five rounds. At least one argument in his favor can be based on how you read 29.E.4. I rejected using 29.E.4 to force a switch, but an argument can be made to use it. In the absence of using 29.E.4, the 3-point difference for both switches means that either one can be used and still have the pairings be legal, so with both being viable you have your option as to which one to choose.

11

I see people talking about 3 point rating differences. Maybe it’s just meant to highlight how close the ratings are. But relative to any other consideration (e.g. color), it seems to me that a rating difference of 10 or 20 points is unimportant.

Do people here agree with me? I’m a former LTD from pre-computer days.

12

Depends whom you ask, I guess. Jeff apparently is using small rating differences to tie-break two possible pairings which he regards as equally valid. I’m more concerned about colors, going beyond mere equalization and alternation.

Apparently the TD, too, was concerned about assigning BBWBB:

– but overlooked a third possibility, the one suggested by Jeff and me.

I’m curious – when pairings were made, was this third possibility not thought of, or was it thought of and rejected, and if it was rejected, why? Of course, I realize how easy it would be to overlook a possibility, especially with so many rounds and so much trickiness involved in preventing two players from being paired against each other twice, etc.

I’m also curious – what would various pairing programs have done here?

Bill Smythe

13

WinTD produced Jeff’s pairing for boards 1-5.

14

I’m happy we could be of service providing fodder for your ANTD and NTD exams! :slight_smile:

Bill Smythe

15

Well, these are the actual SwissSys pairings. We decided to go with them. The chief TD’s rationale was that Ehlvest had had WB WB BW B, so it was really only a second Black in a row after alternation. Since we did pairings immediately after the round ended, we had plenty of time to look them over, and even with bringing in a special referee, we gave the players sixteen hours to prepare.

As far as I can understand, then, the special referee felt no great time pressure, and he paired the scoregroup himself, getting the same results that the computer did. I don’t know if he was using the computer results as a guide or not. He did mention that he felt that there was no rule considering an internal sequence of colors in the middle of a tournament, i.e. that as long as a player had to get a fifth Black in eight rounds, it wasn’t important to look at partial color histories, except to see that he hadn’t had Black in the previous two rounds. I hope that I am fairly representing his position.

Alex Relyea

16

Thanks for the insights. I guess I’m almost ready to stop beating this not-quite-dead horse.

It appears there were four reasonable alternatives:

A. Treat Gulko as the odd player. In the 4.5 group, resolve the already-played problem (Ehlvest-Perelshteyn) by transposing Perelshteyn with Ivanov.

B. Treat Gulko as the odd player. In the 4.5 group, resolve the already-played problem (Ehlvest-Perelshteyn) by transposing Perelshteyn with Becerra.

C. Treat Perelshteyn as the odd player. This leaves no already-played problems in the 4.5 group.

D. Treat Becerra as the odd player. This leaves no already-played problems in the 4.5 group.

It appears that, during the pairing discussions among the TD staff, only options B and D were considered:

All four options assign the black pieces to one player due white. There’s no way around that. But:

Advantages of A:

  1. Reasonable odd-man choice.
  2. Nobody gets 4 blacks in 5 rounds.
  3. Win-TD thought of it.
  4. Jeff thought of it. :slight_smile:
  5. I thought of it. :slight_smile: :slight_smile:

Advantages of B:

  1. Reasonable odd-man choice.
  2. Swis-Sys thought of it.

Advantages of C:

  1. No transpositions required in the 4.5 group.
  2. Nobody gets 4 blacks in 5 rounds.

Advantages of D:

  1. No transpositions required in the 4.5 group.
  2. Nobody gets 4 blacks in 5 rounds.
  3. The TD staff considered it. :slight_smile:

Disadvantages of A:

  1. A transposition is required in the 4.5 group.
  2. The TD staff didn’t think of it. :confused:

Disadvantages of B:

  1. A transposition is required in the 4.5 group.
  2. One player gets BBWBB.

Disadvantages of C:

  1. Questionable odd-man choice.
  2. The TD staff didn’t think of it. :slight_smile:

Disadvantages of D:

  1. Questionable odd-man choice.

Bill Smythe