Tipster rating
Every pick by an author enters the statistics, win or lose. Next to the return you can see the sample size, the accuracy and the CLV, so that working the line can be told apart from luck.
TheAnalyticK model
model
Yellow cards
85
76.5%
+10.94%
+18.74%
+1.11%
2
TheAnalyticF model
model
Match result
193
50.8%
+5.05%
+18.92%
+1.48%
3
TheAnalyticC model
model
Corner totals
249
55.8%
+0.06%
+0.28%
+1.23%
4
TheAnalyticT model
model
Goal totals
518
54.6%
-0.01%
-0.06%
+1.76%
5
TheAnalytic model
model
All leagues · rating model
68
45.6%
-1.37%
-1.66%
+1.64%
6
TheAnalyticX model
model
Match result
294
42.2%
-4.34%
-22.47%
+1.14%
7
TheAnalyticD model
model
All leagues
858
45.1%
-5.7%
-97.76%
+0.8%
What CLV is
The difference between the odds the author locked at publication and the odds at the close of the line. A positive CLV means the author is systematically ahead of the market rather than guessing individual matches right.
Minimum for the rating
Over 30 days — no fewer than 10 settled picks and average odds from 1.3. The odds threshold rules out building statistics on near-certain outcomes.
Part of the models’ history was published retroactively: the verifier arrived later than they began computing. Such picks do count towards the rating — their estimate is pre-match and the odds are taken from the opening line. The grounds differ, though: for the TheAnalytic model the computation time is recorded in the source and is verified, while for TheAnalyticD the source does not store it and the pre-match nature is assumed from how it works. In the profile and on the pick card this is labelled separately.
ROI and the bankroll answer different questions
ROI — how much each unit staked returns; it measures selection quality and does not depend on how much the author staked. Bankroll — how much the starting bankroll grew if you followed the author in full. A high ROI with a modest bankroll means rare but accurate picks; the opposite means many bets with a thin edge.
How the order is built
Rows are ordered by ROI trimmed by sample size: while there are fewer picks than 75, the return counts in proportion to their number. This is not an adjustment to the return itself — the table shows it as it is — but a safeguard against +40% over a dozen picks ranking above +8% over three hundred. The former is nearly always luck, and on a short sample that is indistinguishable from skill. Calculation method →