Predictions 2014-2015

[J.E.]

I get 2 different winners, depending on how you score it

Code: Select all

╔════════════════╦═══════════════╦══════════════╦══════════════════╦═════════════════╗
║                ║ PythWins RMSE ║ PythWins ABS ║ Actual Wins RMSE ║ Actual Wins ABS ║
╠════════════════╬═══════════════╬══════════════╬══════════════════╬═════════════════╣
║ 538            ║ 7.79***       ║ 6.10***      ║ 8.16             ║ 6.20***         ║
║ AcrossTheCourt ║ 8.49          ║ 6.33         ║ 8.89             ║ 7.09            ║
║ AJ ASPM        ║ 9.11          ║ 7.24         ║ 9.82             ║ 7.80            ║
║ AJ PT PM       ║ 7.73**        ║ 6.00**       ║ 7.82*            ║ 6.03**          ║
║ bbstats        ║ 7.57*         ║ 5.96*        ║ 8.01**           ║ 6.00*           ║
║ Bobbofitos     ║ 9.17          ║ 7.24         ║ 9.64             ║ 7.93            ║
║ Crow           ║ 8.22          ║ 6.57         ║ 8.09***          ║ 6.27            ║
║ Dr Positivity  ║ 9.02          ║ 6.84         ║ 8.80             ║ 6.73            ║
║ ESPN Fall      ║ 9.02          ║ 6.84         ║ 9.09             ║ 6.80            ║
║ ESPN Summer    ║ 8.95          ║ 6.84         ║ 9.06             ║ 6.60            ║
║ fpliii         ║ 9.39          ║ 7.32         ║ 9.11             ║ 6.99            ║
║ HoopDon        ║ 8.44          ║ 6.36         ║ 8.63             ║ 6.60            ║
║ italia13calcio ║ 8.50          ║ 7.07         ║ 8.40             ║ 6.90            ║
║ Mike G         ║ 9.31          ║ 7.52         ║ 10.05            ║ 8.15            ║
║ mystic         ║ 8.16          ║ 6.43         ║ 8.40             ║ 6.34            ║
║ nbacouchside   ║ 8.84          ║ 6.90         ║ 8.68             ║ 6.67            ║
║ sndesai1       ║ 9.84          ║ 7.32         ║ 9.70             ║ 7.17            ║
║ v-zero         ║ 8.64          ║ 6.95         ║ 9.14             ║ 7.30            ║
║ William Hill   ║ 9.01          ║ 6.98         ║ 9.06             ║ 6.95            ║
╚════════════════╩═══════════════╩══════════════╩══════════════════╩═════════════════╝

I'm not sure whether official scoring rules were stated before the season, but I think we can safely say that bbstats and AJ PT-PM did best, with 538 not far behind

[sndesai1]

Thanks JE. I think there were a few entries added halfway and I ended up changing the Vegas line to the closing LVH one so those be the only differences between your 3rd column and chart below:

[permaximum]

Congrats AJ.

So RPM or xRAPM - so-called the best metric - lost again to something not even a blend

[bbstats]

Well...mine was built to predict RPM in the same way PT-PM was built to predict RAPM, so....



Thanks guys, that was fun.

[EvanZ]

So when do we get PT-PM predictions for the playoffs?

[permaximum]

Well...mine was built to predict RPM in the same way PT-PM was built to predict RAPM, so....



Thanks guys, that was fun.

Hmm, I thought you used the RPM itself and I was a bit surprised with the end result. Now it makes sense

[Mike G]

Does anyone feel like digging up previous years' predictions and see what are correlations from year to year, from various sources?

Last year's winner was near the bottom this year. Just saying.

[Crow]

Congrats to the winners.

Too much weight to Vegas this season when Vegas was meh. At least my weighting was enough of a contribution to beat 538 compared to simple averaging.

search.php?keywords=predictions+&terms= ... mit=Search

The 2009-10 data is missing. 2007-8 only briefly referenced. Nor sure if there was a contest before 2007-8, maybe one or two more happened before that. So it is a little cloudy. Hollinger and I each have 2 wins since 2007-8. Neither of us competed every year, but I competed more. I have at least four top 3 finishes in I think 5-6 tries in the time period; he had at least 3. No other multiple winners yet, unless it was Vegas in the early days.

[Mike G]

For 6 peeps in last year's and this year's predictions, the correlation is .553
If we take out Bobbofitos, it's .924 -- the rest of us are in the same order each year.
If we take out yours truly, it's -.186

[Crow]

It might be interesting / useful to backwards engineer an optimal blend. It will change year to year but if I was serious about trying to win again I should do it for a period of years more precisely instead of just eyeballing it and roughly blending. The competition appears to have gotten tougher.

[Bobbofitos]

Does anyone feel like digging up previous years' predictions and see what are correlations from year to year, from various sources?

Last year's winner was near the bottom this year. Just saying.

don't remind me