Validating tennis probabilities

Betting statistics

select SUM(total) from bets
select sum(total) from (select *, row_number() over (partition by event_id,selection_id order by latest_taken desc) as row from bets) A where row=1

-- P&L per weeek
select row_number() over (order by DATEPART(year,settled_date),DATEPART(wk,settled_date)),DATEPART(year,settled_date),DATEPART(wk,settled_date) as week
,SUM(total) as total from bets group by DATEPART(year,settled_date),DATEPART(wk,settled_date) 

--P&L per probability range
select cast((1/odds)*100 as int) as price_implied_prob,sum(total) as profit,count(*) as num_of_markets, SUM(cast(win_flag as int)) as wins,
100*SUM(cast(win_flag as real)) / count(*)  as true_prob, 100*avg(predicted_prob) as predicted_prob
from bets s group by cast((1/odds)*100 as int) order by price_implied_prob 

--Reload probabilities
delete from market_prob
--insert market probabilities to sql
BULK
INSERT market_prob
FROM 'C:\daniel\betfair_data\tennis_prob_data.csv'
WITH
(
FIELDTERMINATOR = ',',
ROWTERMINATOR = '\n',
FIRSTROW=2
)
delete from market_prob where probability=0