Skylark2009
Colt
Dr. StrangeLAY Or: How I Learned to Stop Worrying About the Winners and Embrace the LosersThis thread documents a pure laying system: selections + results only.
No reasoning, no form reading given, no narrative — just a structured treasure hunt.
All bets are placed automatically via Gruss Betting Assistant linked to an Excel sheet containing the selections.
This is primarily about the Kelly based staking and the fine tuning of the fractions that are being applied.
This thread will move very slowly at first, but may pick up if the bank grows.
Core PrinciplesSelections are posted daily before the days racing (no edits to posts with selections), any errors I find after posting means I will delete and repost.
Results are posted daily (mostly).
No explanations for selections — this is intentional.
All staking is automated using a Kelly‑based model via Gruss/Excel.
Bet Placement RulesExchange used: Betfair
Odds range: 1.01 to 1000.
Timing: bets placed about- ~15 seconds before each race's listed start time.
Market: bets to be placed into the BSP market.
N.B. there will be differences to the calculated Kelly liabilities/returns due to BSP not being known when the bet is placed.
Liquidity: bets not matched due to no market liquidity before the race actually starts they will automatically cancel or lapse or not be placed at all.
Bank & Risk ManagementA portion of my Betfair balance is allocated to the system bank (£200.00).
If the bank falls below £25.00, I'll consider ending the thread as recovery at this level would be very slow.
Expected Return: Modest but consistent. The strategy relies on gradual growth, with compounding increasing the returns over time.
High-Odds Selections: These generally produce very small returns and may occasionally remain unmatched due to limited liquidity.
Losing Runs: The bank/staking structure should withstand roughly 8 to 12 major losing liabilities close together before being compromised.
500pts will be withdrawn each time the bank passes 2500pts. This should keep the liabilities and returns in check.
Kelly-Based Lay Staking ModelThe staking model uses a modified version of the Kelly Criterion for laying selections (fractional Kelly).
The calculation is based on:
The market-implied probability of the selection losing.
A user-defined edge.
The current bank size.
This should generate a positive percentage of the bank which will be used as the liability (not returns).
Fixed EdgeTo ensure every selection qualifies for a bet, a very small edge is added to the market-implied probability of losing.
Example
Bank equals 200pts
BSP 10, fractional 9/1 or 10% implied probability to win
Market implied probability of losing: 90.0000%
Added edge: 0.0025%
Effective implied probability to lose: 90.0025%
This creates a small positive Kelly value, allowing a stake to be generated automatically, in this case, 2.5% or 5pts from the current bank as liability.
Kelly Variation Applied by odds band (fractional Kelly)Each calculated Kelly% will be further reduced by multiplying by the values in the 'Kelly fraction' column depending each selections BSP.
Min BSP | Max BSP | Range | Kelly fraction |
|---|---|---|---|
1 | 2 | <=2 | 0.5 |
2 | 10 | >2 <=10 | 0.5 |
10 | 20 | >10 <=20 | 0.75 |
20 | 30 | >20 <=30 | 0.75 |
30 | 40 | >30 <=40 | 0.25 |
40 | 50 | >40 <=50 | 0.25 |
50 | 100 | >50 <100 | 0.25 |
100 | 400 | >=100 <400 | 0.1 |
400 | 1000 | >=400 | 0.1 |
Caveats -This thread will produce high priced losing lay bets!
If no suitable liquidity exists when the bet is due to be placed, the bet will not be submitted by Gruss resulting in an NR in the results.
Non-bet selections will be recorded as NR (No Result), covering unmatched bets, insufficient liquidity, withdrawals, non runners, voided races, abandoned meetings or similar circumstances.
Betfair commission of 2% is only being applied in the summary tables after the results are known for informational purposes. The ongoing liabilities (Liability), bank (Bank) and running totals (Rtot) columns in the results do not reflect this and are commission free while moving forward.
Where multiple selections occur in the same race, each selection's liability is calculated using the same pre-race bank value.
I'll likely be adjusting the Kelly fractions, withdrawal rates, the added fixed edge I started with etc, as the thread progresses.
