Ed Wheatcroft

[Leonard Smith @lynyrdsmyth]

Pi in the sky!

Happy pi day.

[Ed Wheatcroft @edwheatcroft1]

@LSEStatistics @LSEnews @SocialStats_RSS Devastated to hear this news. Henry was a brilliant mentor to me and countless others who were lucky enough to work with him. But most importantly of all, he was a fantastic friend. He will be sorely missed.

[Leonard Smith @lynyrdsmyth]

New academic year. New #NFL season tonight. New Home-win Forecasting App.
Working with #VT students to practice decision making and uncertainty guidance, as they feel the risk of forecasting the probability of a how-win for #Chiefs in #BALvKC
@VTECE @edwheatcroft1 @sdac_vt

[Leonard Smith @lynyrdsmyth]

Week 16 #NFL Home-Win #Probability #Forecasts

@cfrelund outside (more confident than) @oddschecker in #LVvsPIT !

[Ed Wheatcroft @edwheatcroft1]

@lynyrdsmyth @RamsNFL Knowing very little about the teams, I would have to go with the bookmaker's implied probability on the @RamsNFL: 0.26.

[Leonard Smith @lynyrdsmyth]

Probability forecasts for Week 13 of the 2022 @NFL Season. Six lone wolf forecasts is a record hi.

Meteorologists (& other modellers) might find it useful to make regular sports related probability forecasts. Competing with 5 skilful forecasters every week has taught me a lot!

[Leonard Smith @lynyrdsmyth]

Home-Win #Probably #Forecasts: 2022 #NFL Week 10
Only two games with agreed heavy favourites: #JAXvsKC and #WASvsPHI .
@cfrelund a lone wolf: @AZCardinals in #ARIvsLAR.
@ESPNStatsInfo model finally draws a purple light for pre-game model instability, not for its low score. Why?

[Ed Wheatcroft @edwheatcroft1]

@lynyrdsmyth @cfrelund @AZCardinals @ESPNStatsInfo @FiveThirtyEight @oddschecker @NFL @espn @ESPNStatsInfo pre-match forecast probs are showing large fluctuations leading up to the game, presumably based on assumed player availability. This effect is much larger than in betting markets, suggesting overreaction to this info.

[Ed Wheatcroft @edwheatcroft1]

@lynyrdsmyth The value of a set of forecasts has to depend on a the nature of the thing you're forecasting. This set of forecasters could be insightful (i.e. they know the game is a toss up) or just know very little. We can only tell by evaluating their forecasts over multiple games.

[Leonard Smith @lynyrdsmyth]

Home-Win #Probability #Forecast for Week 9 of the 2022 #NFL Season.
#LVvsJAX called closest, while @ESPNStatsInfo and @cfrelund usually confident on #TENvsKC and #BUFvsNYJ respectively.
All forecasters favour @BuffaloBills and @Chiefs with a geometric average near 80%.

[Ed Wheatcroft @edwheatcroft1]

@lynyrdsmyth If the 'best' forecaster has a high chance of 'losing' even over a full season, how can we discourage them from gambling by making overconfident forecasts that don't reflect their true beliefs?

[Leonard Smith @lynyrdsmyth]

Home-Win #Probability #Forecasts for 2022 #NFL Week 7

@cfrelund lone wolf in #NYGvsJAX , now “Called Closest”
Strong agreement on @Buccaneers as last week, how does this inform #TBvsCAR?

Why did I move to 0.49 in #NYJvsDEN given my true belief is 0.50 and IGN a proper score?

[Leonard Smith @lynyrdsmyth]

Predicting the Predictors:
Will be interesting to see the probability forecasts of those near the top of the 538 leaderboard tonight for #NFL #WASvsCHI game.
I’m forecasting 50% & predict Ross will be near me with Shawn farther & Mark farther still.
Why? (538 use a proper score)

[Leonard Smith @lynyrdsmyth]

While I adore physics based violations of Cromwell’s Rule (“Never say zero probability in life.”), this claim suggests you review your scorigami model:

Hou TD for 7 pts: 20 - 6
JAX FG for 3 pts: 20 - 9
JAX kickoff muffed,
-> Saftey: 20 - 11.

>3 min; small, nonzero prob.

[Leonard Smith @lynyrdsmyth]

2022 #NFL Week 5 Home Win Probability Forecasts

[Leonard Smith @lynyrdsmyth]

Home-win #Probability #Forecasts for #NFL Week 3.
#LVvsTEN called closest game.
Four Lonewolf games including @cfrelund in #GBvsTB , she remains significantly ahead of all others even though noone lost points on #TNF and most reduced the gap.

[Leonard Smith @lynyrdsmyth]

Home-win #Probability #Forecasts for #NFL Week 2
#LACvsKC tonight & each forecaster on our list favors the @Chiefs over @chargers in #ChiefsvsChargers.
What does the alignment in these probabilistic predictions really tell us?
#statistics #maths #DataAnalytics #DataScientists

[Leonard Smith @lynyrdsmyth]

Home-win #Probability #Forecast: #NFL Week 1
Favorite Unclear in 4 games, #CLEvsCAR Called Closest.
One Lone Wolf game #GBvsMIN
Forecaster's probability high on @Colts in #INDvsHOU; insight or week 1 over-confidence again?
Model Inadequacy flag on Elo #DENvsSEA, a silly input?

[Leonard Smith @lynyrdsmyth]

Cromwell’s Rule forbids assigning a probability of zero to anything that is not a mathematical impossibility, but what is a physicist to believe when, with only 41 seconds remain in the game, the team trailing by 21 points throws an incomplete pass on fourth down?