Bankroll and Variance
Survive long enough for edge to show. Sizing and drawdown tolerance are part of the model — not an afterthought.
Kos Edge
Loading desk…
Pulling the latest board. This should only take a moment.
Insights · House rules
Evergreen process pillars. How KosEdge thinks — free by design. No module numbers. Trust the desk before you pay for the diary.
10 pillars · updated for the live desk · no public module numbers
Survive long enough for edge to show. Sizing and drawdown tolerance are part of the model — not an afterthought.
Closing line value is a useful lie detector for process — not the whole religion. Beat the close on average; don't worship every tick.
Distributions become lines; lines become decisions. Simulation is not a pick — it is the machinery that produces a fair you can bet against.
Official > trusted beat/insider > noise. Reprice only when the information earns the right to move your number.
Independent fair comes before market respect. We price the game, then we look at the board — never the other way around.
Research fair, handicap reprice, edge only vs market. Three layers — don't collapse them into one number.
An empty slate is allowed. The desk does not owe you a card. Manufactured bets are how edges disappear.
Key numbers, juice, and thresholds decide whether a "good side" is actually a good bet. Side without price is just a rooting interest.
Grade good bet / bad bet by price and process, not by the scoreboard. Results are lagged, noisy feedback.
Pass is a position. There is no "close enough." If it doesn't clear the threshold, it doesn't clear the board.