Thread by @singhpr

1/n In-match, win probabilities for the #IndVsPak world cup game yesterday. This was a nail-biter of a game with a huge swing in fortunes towards the end. The win probability forecaster (by @ESPNcricinfo) has some lessons in #ContinuousForecasting
#metrics #agile

2/n The forecast is probabilistic, not deterministic. It always (except at the end..more on that in a bit) talks about the probability of an outcome, not "This shall happen". It is not a single - "Ind will win" or "Pak will win".

3/n How about your forecasts? Are they of the type - "I will get this done in 5 days" or "I have an 80% chance of getting this done in 5 days"

3/n At no point does the forecaster say 100% till the end of the game. There is a point towards the end where Pakistan is projected to win with 90% confidence, but that is as high as it goes before there is an actual winner.

4/n A vast majority of events (weather, sports, sprint planning, release planning etc.) follow this principle - We can very rarely be 100% certain of an outcome. The only time we can be certain is after the event has happened. There are always multiple outcomes possible.

5/n As the game progresses, the predictor re-forecasts. India is favoured to win at the beginning of the game, but as the game progresses, you can see the advantage being squandered. To the point that they had to resort to end-of-game heroics.

6/n How often do you reforecast? Also, how often do you react to forecasts? Do you let things slip till you need "end-of-game" heroics? Or do you make early interventions when the advantage starts slipping away?

7/n Re-forecasting is improving your predictions as new information becomes available. Every play, every delivery, every shot at goal, every story finished is information that can help us understand our current situation better.

8/n This graph shows that Pakistan managed the game really well and still lost. India, on the other hand, won a horribly managed game. that is just how probabilities work.

9/n The point of forecasting is to understand the probabilities and play the odds by reacting to the predictions. India did that well right at the end of the game, having not done it well throughout the game.

Want to learn more about forecasting for your teams - Check out @prokanban's Applying Metrics for Predictability public courses - prokanban.org/applying-metri…
Reach out to me or to [e-mail] if you are looking to set up a private class.

@threadreaderapp unroll