Yank model: Challenger to the Duckworth/Lewis mode

[AbhishekS]

D/L is the best system

Not for T20s IMO.

 

[pcfan123]

I think that if a game is not played 20 overs a side then it should not even be a game.

5 overs a side is a tweaking joke, then again so is 2020

 

[yashsr]

well, I like the idea, and your projections certainly look better a lot of the time.

however, bit of creative maths on the T20 projected scores.

in a few cases D/L has under projected scores and yours has over projected them. This has led to you saying, for an actual score of 150 where YANK has given 166 and D/L has given 134, "YANK is -16" and "D/L is 16"

that's fair enough, however, when you've totalled the differences you've used the under projections as positives and the under projections negatives. This has meant, the D/L, which more often under projects than your system was out by 309 and your system which over-projects a little more is only out by 92. Doing it this way the YANK looks massively superior.

but imo, the above example of YANK over-projected by 16 and D/L under-projecting by 16 is all one and the same. It's still wrong by 16.

I did the calculations again, taking out the negatives and found overall, your system to be out by 502 runs overall and D/L out by 523. This means your system is on average wrong by 22.8 runs a match, and D/L is out by 23.7 Runs a match.

However, far more worrying, the standard deviation for D/L is actually lower than yours

YANK Standard Deviation = 28.19
D/L Standard Deviation = 24.62

so actually, D/L might under-project a bit more but is more consistently returning closer scores.

However I will grant you that your system has been more harshly punished by the extreme data in the table (the wayamba collapse game) and taking out all scores where both systems were 30+ out they even out at 19.

Still doesn't make your system that much better in T20, and the maths you've used to suggest it is either deliberately misleading, or just bad.

Thanks for your feedback, Yes you're correct, we haven't put up absolute totals. The thing we're trying to prove with this simulation is that D/L almost always gives low projected scores which means its acceleration level for T20 is very very low as compared to what happens in an actual game.

It is really impossible for a model to come very close to real match scores because of matches like:
Match 17) South Australia scored 191 from a relatively bad position of 68/3 in 10 overs

Match 16) Wayamba score 106 from a very healthy start of 81/2 in 10 overs

When we take normal differences(not absolute), we can find out whether the acceleration level of the model is simulating what happens in real matches. We initially planned to take absolute differences but it wouldn't prove any point because of bizarre matches.

Readers are actually requested to look at the 10 over score of the team and project the 20 over score for their team according to their own judgement and then have a look at Yank and D/L Model Projected Scores rather than only comparing with the actual score attained in the match. Using the actual score of every match wouldn't be the best way to judge, using a total of actual score would a far better measure because of extreme matches....

@others: Thanks for your feedback...yes we'll keep you updated If there is any match you think are worth covering, let us know....

yashsr added 11 Minutes and 26 Seconds later...

That is something I never knew. And will be the one thing that I learn today, I still like the way you have gone about your testing and designing of the algorithm and tested it out so far.

Out of interest I saw that you had been tweeting it to cricket players, past-players & pundits, how did they respond?

We aren't getting a lot of responses because many celebrities don't even read every message they get on twitter.... But we did hear back from Sanjay Manjrekar, Akash Chopra and Srinivas Bhogle(brother of Harsha Bhogle...Mr. Srinivas has written numerous articles on D/L and VJD earlier)

 

[Ollie_H]

What comments did you get from them about your findings?

 

[yashsr]

What comments did you get from them about your findings?

They all said: 'Interesting will check it out!'....

But Srinivas Bhogle is in constant touch with us and he is guiding us really well...This is perhaps because he is the only person who I think would be able to understand the method pretty well and quickly too because he's done a lot of research on D/L and VJD

We also heard back from David Llyod and Pommie Mbangwa this week...but they're yet to study it carefully

 

[sohum]

This is kinda nifty. Will definitely check it out when I have some time.

 

[Biggy]

D/L is the best system

Yes, take the first ODI between Bangladesh and New Zealand this month.:sarcasm Surely there's better.

 

[yashsr]

I think we've focused a little to much on the Real Match Simulation...we'd love to have feedback on the general performance of the Yank Model as against D/L model and VJD model both in ODIs and especially also in T20..

 

[yashsr]

About yesterday's match which was the 2nd ODI in the series between SL and Aus:

SL 161/2 in 34 overs.
Match reduced to 45 overs.
SL 213/3 in 41.1 overs.
Innings ends and Aus get only 39 overs to bat.

39 over Target

Yank Method :240

D/L Method :244

VJD Method :231

Aus 139/5 in 27.4 overs
Match reduced to 38 overs.
Find target for Australia in 38 overs.

38 over Target

Yank Method :236

D/L Method :240

VJD Method :227

 

[Owzat]

Erm, how do you prove which adjusted target was right? I mean if the side batting second reaches the target does that mean it's right? Or does whether they reach the target or not depend on how well they bat?

For me a problem within a lot of the chases is they are so public. Sides know what they need to do, it becomes the focus of the game rather than winning it. While sides know what they need to do it creates an artificial game, and not necessarily a good one. What they really need to do is make a realistic adjustment based on where the side chasing would be if the lost overs had been batted through without further loss.

So for instance chasing 260 off 50 overs the side chasing is 50/2 after 10 overs and loses 20 overs, it is fair to assume they'd score something between 3 and 4 runs per over if they didn't lose another wicket leaving a chase off the last 20 overs of 130-150. I'm sure most sides/people would accept that as a not unreasonable assertion, the toss would most likely be argued over the exact number of runs, but since it would only add up to a maximum difference of a run an over then it wouldn't be that crucial. Maybe they could roll a die (1-20) and that would be a random factor that could influence any match anyway.

I hate Duckworth Lewis as it ruins a good game of cricket, even when it isn't raining you can get commentators clutching to their DL sheet citing where sides should be if it rained just to gauge the chase.

 

[yashsr]

Erm, how do you prove which adjusted target was right? I mean if the side batting second reaches the target does that mean it's right? Or does whether they reach the target or not depend on how well they bat?

For me a problem within a lot of the chases is they are so public. Sides know what they need to do, it becomes the focus of the game rather than winning it. While sides know what they need to do it creates an artificial game, and not necessarily a good one. What they really need to do is make a realistic adjustment based on where the side chasing would be if the lost overs had been batted through without further loss.

So for instance chasing 260 off 50 overs the side chasing is 50/2 after 10 overs and loses 20 overs, it is fair to assume they'd score something between 3 and 4 runs per over if they didn't lose another wicket leaving a chase off the last 20 overs of 130-150. I'm sure most sides/people would accept that as a not unreasonable assertion, the toss would most likely be argued over the exact number of runs, but since it would only add up to a maximum difference of a run an over then it wouldn't be that crucial. Maybe they could roll a die (1-20) and that would be a random factor that could influence any match anyway.

I hate Duckworth Lewis as it ruins a good game of cricket, even when it isn't raining you can get commentators clutching to their DL sheet citing where sides should be if it rained just to gauge the chase.

For your information, commentators do not use D/L sheets anymore because the current D/L model requires a use of the computer...They used sheets way back until I think 2004 when D/L had its standard version...

I think you'll find the computation part of the Yank Model and the D/L model very interesting..The technical part of the Yank Model will be put online when we've completed our patenting/copyrighting process...You cannot roll a die to take into account the run-rate, it has to be more scientific...

When there is an interruption(whether single or multile) in the 1st innings, Yank Model finds its 50 over par score(also known as projected score) and then finds the 2nd innings target accordingly. This is all I can say for now!!

 

[Owzat]

But you still haven't answered the question? If DL produces a different score to the yank method, how do you know which is right? (or more right?) You don't. All you can do is make a case for one over the other, might even be better picking a point in the middle!

 

[TumTum]

The technical part of the Yank Model will be put online when we've completed our patenting/copyrighting process

Ok till then stop advertising here. What is the point of all of this if we don't even know how it works?

 

[yashsr]

Ok till then stop advertising here. What is the point of all of this if we don't even know how it works?

I just wanted a feedback as to what people over here think about the performance of the model over the other models... Even if you don't know the technical part, people can still tell us if Yank's targets are better than D/L and VJD or if they're worse...So our objective is to get a feedback.

@Owzat : This is obviously a subjective concept... No target can be accurate, it can only be precise. The target which is the best estimate of a simulation of real match is the precise target.... I believe that we've better simulated the real match situation than either D/L and VJD. But obviously you can differ and can tell us the exact opposite. That is what I posted this thread for. To get feedback and constructive criticism from the members of this forum.