What can you tell from a puzzle's stats?

[midnight]

Here are a few things that I have picked up on:

1. While the boards with a huge number of common and ultra rare words (150+ of each) can be very high scoring, they are seldom the best boards for high-scoring on. The best boards of all for high-scoring generally have around 120+ ultra rare words, and about half as many common words. If you know your anagram and subanagram sets well you can post huge scores on such boards, without wasting too much time typing in easy sets.

2. The boards with such stats almost invariably have a K in a prominent position, usually with at least two Ss and two Es as well. Learn the relevant anagram and subanagram sets well and you are set for some high scores (I have scored a number of 700+ scores on such boards in the last few days).

Boards with very high common and ultra rare word tallies will tend to be made up in the main of fairly common letters. You are most likely to find such things as SERINES or STEANES and their subanagram sets on such boards.

3. I generally estimate what the previous high score divided by the previous highest number of words is. This gives me a good sense of just how much potential the board has for high scores. The best boards will tend to have points to words ratios that are 5 or more. Type in 120 words on a board on which a points to words ratio of about 6 is quite achievable and you have a score over 700.

4. Don't focus too much on how many longer words there are on the board. High scores are easier to make when you aren't trying too hard to find longer words.

5. However, if you see a board on which someone has played a high-scoring or long word and you feel that you need it badly enough to take the lazy route (I do not recommend this), you could Google the name of the player who found it, followed by 'wordtwist' and get to their trophy room (either by clicking on the link if it comes up, or by clicking on their name in a competition results table) and get a pretty good idea of what their word was by looking at the more recent entries in their competition stats, or their awards for longest or highest scoring words. Of course, this is a very poor way to find long words and removes the sense of achievement that you get from finding such words yourself.

6. Pay attention to the number of times that a board has been played. The more times that a board has been played, the less chance there is of your getting the highest score on it. That said, I have beaten previous records on some medium sized boards that have been played over 500 times by well over 100 points on several occasions in the past.

7. Also pay attention to the name of the previous record-holders. I have a sense of how much I can beat most people by and get a sense of the score that I will get by seeing how certain players have scored in the past. Some players' records are particularly hard to beat (snyderman05 is one I find especially hard - I am 90%+ sure that I know him from other word games). When I see someone tough as the current record holder I will often just aim to beat their score (this is generally what I do with my own scores - on occasions I have beaten scores that I set within the last month by over 100 points, a reminder that factors such as form, where you start on the board, etc, can have a significant effect on your performance in a game).

8. Once you have taken all of these things into account, estimate the average score that you would expect to score on such a board. Get a sense of how lucky you feel, how quickly your mind and fingers are working and adjust your estimated score based on such considerations.

Once you click start game, NEVER look at your score racking up, nor at the time. Pretend that they aren't there. You should be able to get a pretty good sense of how you are doing if you are attentive enough to the relative values of certain word sets and how many words you are typing. Glancing at the time or your score will simply distract you. Use the impression that you got of the board from the game stats to get a sense of how well you are playing.


Does anyone else have any tips, or thoughts on what you can learn about a board from looking at its stats?

[GClayman]

First post, though you certainly know who I am, Mr. midnight :-)

My stats are obviously skewed because, with this name, I will only play the biggest boards I can find -- at least 90 common and ultra-rare words for the 4x4, and 100 common/ultra-rare for 5x5.

I agree, though, you can tell from the words and high score what boards are going to be good before you play them. The 5 points-per-word rule is very helpful.

One additional strategy is to play boards that not a lot of people have played. For 5x5 boards, if there are 150 common and 130 ultra-rare words, yet the board has been played 180 times and the high score is only 550 points or so, it's probably not a good board for high scores. Most of the boards I've played have been played less than 50 times or so, which is probably scary when I'm putting up high scores that look to be so-far-out-of-reach to other players. Truth be told, a lot of the boards I'm scoring 700 points on could easily be in the 800-900 point range. I wish there was a feature to see when people beat your high scores.

Long words also mean nothing here. You and I are both players that don't focus on the long words but rather the anagram sets which could be why we average such high scores. Keep in mind the simple 'antes/etna/nates/neats/stane/stean' anagram set is one of the highest scoring with the format used here -- I'm almost certain those are all "ultra-rare" words.

[midnight]

I pay far less attention to the number of common words. I will play the majority of boards with more than 80 ultra-rare words. My general goal is to either raise my overall average score by 1 every session, or, failing that, to leave it no lower than when I started. When I have raised it very quickly I will often play a low-scoring game that has been played 700 times or more and try to beat the top score, just for the change of pace. Occasionally, if I see a familiar name holding a record in a game that has been played many times I will play the game just to beat the person, even though I know that the board will probably be low-scoring.

I have toyed with the idea of starting lists in which I set Wordtwist point values to the most common SOWPODS anagram sets. It might prove useful to know where to concentrate one's attention when learning such lists.

[alicat]

Hi Guys :)

Midnight, I can tell pretty much the same as you - I can score highest on boards with many (near 100) common and ultra rare words. However, I tend to score highest on boards with many common words, particularly those with a large number of 3-letter common words (>25-30).

At a glance, a full grid (words for all number of letters in all categories) tends to indicate many words, though not necessarily the highest scoring.

As far as who's already played - this one is slightly less reliable. Midnight, if you've played, I know that my score is likely to be near yours, and that I probably wont score as high as you, or at least not much higher than you. CG, I usually cannot score as high as you. There are a few of other players who also fall into this category. Not that I will necessarily score higher than the highest player if it is anyone else, but I am usually not able to predict other players as well.

Of course the same is true for me regarding the number of times played; if the board has been played more than 300-400 times, the standing high score is typically near the maximum, or what I will score. Of course there are exceptions, and I have also beaten high scores on well-played boards.

One trick I have picked up that neither of you have mentioned yet (although you have alluded to it, Midnight): if a board's high scores dont seem to match up with its statistics (i.e. a board with 100+ common and ultra-rare words that has a low high score, even after several plays), I have found that if I rotate the board 1-2 times before starting to play I can often score much higher than the existing high score. This is probably because starting point can have an effect on how well the board plays, as you suggested, Midnight.

[midnight]

Yes, starting point is a huge issue. I have played games on which I hold the high score (and which I have since forgotten) and beaten my previous score by over 150 points. I suspect that the vast majority of this difference can be put down to differing starting points.