Most youd risk on one market

[foxwood]

There's a film in there somewhere

[Alpha322]

Well, I thought it was anyway

Seriously though, would love to hear others thoughts or choices and decisions about how they choose their staking levels.

People especially New Traders think the more bigger your bank the more you can put in and the more you can make. Me personally i use to fit stakes to a price ie if i wanted £5 per tick etc , but i dont do this any more. Your returns are reliant on how well you manage your positions and how much is in the market for you to get out, if i see roughly there is £200-300 stakes on the ladder my biggest stake for that market would be say £500 (Built up of £125 stakes) Scale in, and then scale out if market going for or against. Its about managing and keeping in flow with what the market has with in it not Staking

[stueytrader]

Well, I thought it was anyway

Seriously though, would love to hear others thoughts or choices and decisions about how they choose their staking levels.

People especially New Traders think the more bigger your bank the more you can put in and the more you can make. Me personally i use to fit stakes to a price ie if i wanted £5 per tick etc , but i dont do this any more. Your returns are reliant on how well you manage your positions and how much is in the market for you to get out, if i see roughly there is £200-300 stakes on the ladder my biggest stake for that market would be say £500 (Built up of £125 stakes) Scale in, and then scale out if market going for or against. Its about managing and keeping in flow with what the market has with in it not Staking

Yes, it's a fair point that picking a 'staking level' is not the entire picture of course. However, it's still possible to either scale up, or scale down, in your basic staking amounts used - I'm pretty sure Peter uses far bigger stakes than I do for example, even though we may be using the same markets and/or method to do it.

[stueytrader]

Care to share more detail on this?

I was working for a firm called DST on a trading system called Impart, something like 45 of the top 50 investment management firms used it, front and back office stuff. At the time it didn't have a derivatives trading module. Barings and a couple of others (Nomura and Goldman's I think) sponsored the development of one and I designed it and oversaw the dev. I met Nick briefly when I went out to Hong Kong for the implementation of the first release. A lifetime ago when I was a young hotshot. It was at a time when you could just write stuff and go and install it, when dev got serious and procedural it took all the fun out of it.

Oddly, nerdily, cool to have been a part of the story though Shaun

Did he mention you in his book I wonder....

From sounds of it, he was going against having the kind of safeguards that would have limited his own trading?

[matekus]

Trader Risk-of-Ruin Max Stake Sizing

For traders, here are two ways to calculate our maximum stake per market, held fixed for the duration of a session. Treat the output ('s_max') as a ceiling to stay comfortably underneath, not a target to reach.

The equation we will use is:

'Fixed-Stakes Exponential RoR Approximation'

s_max = 2m * D / (v * ln(1/r))

where:

• s_max = maximum liability per market, as a fraction of starting bank (ceiling)

• m = mean of market outcomes

• D = drawdown from your starting balance (not the running peak - see caveats)

• v = variance of market outcomes

• r = risk-of-ruin

's_max' (in GBP) is held fixed for a session; recalculated at session end.

---

---

The two approaches are just two ways of estimating the historical mean ('m') and variance ('v') of our trading. Session length can be a day's worth of markets, a week's worth, or more. The only constraint is that 's_max' is recalculated at the end of each session, not with each new market traded. Note that 'session' is our choice rather than something the equation provides - see caveat 2.

For both approaches, we have the following common inputs:
- Session Bankroll: GBP10,000
- Drawdown Threshold: 50% (D)
- Risk of Ruin: 5% (r)


**Method 1: Wins and Losses**

If we want to ballpark 's_max' then we need the historical number of markets traded over multiple sessions, the number of profitable markets, and the average decimal odds:
- Historical Markets: 300 # Not meant to be realistic
- Historical Profitable: 156 # Not meant to be realistic
- Historical Average Odds: 2.12 # Not meant to be realistic

At decimal odds of 2.12, a win returns 2.12 - 1 = 1.12 per unit staked and a loss returns -1.00, with p = 156/300 = 52%:
- WL Results Mean (m) = 0.52(1.12) - 0.48(1.00) = 0.10240
- WL Results Variance (v) = 1.12180

From these we calculate an optimistic 's_max', which stake value (GBP305) becomes our fixed ceiling for each market this session:
- WL Max Stake: 3.05% = GBP305


**Method 2: Profit and Loss**

By contrast, if we have access to our profit-and-loss records across multiple historical sessions, then for every settled market we take our net P&L after commission and divide it by the maximum liability we had on that market. That gives a list of numbers such as +0.35, -0.35 and -3.00, each a multiple of that market's liability. For this toy distribution, we calculate 'm' and 'v':
- PnL Markets: [52.00%, 47.50%, 0.50%] # Must sum to 100%
- PnL Results: [+0.35, -0.35, -3.00] # Multiples of liability
- PnL Results Mean: 0.00075
- PnL Results Variance: 0.16689

From these we calculate a more realistic 's_max', which stake value (GBP15) becomes our fixed ceiling for each market this session:
- PnL Max Stake: 0.15% = GBP15

**Twentyfold Difference**

- Counting wins and losses m = 0.10240, v = 1.12180, 3.05% GBP305
- What actually happened (P&L) m = 0.00075, v = 0.16689, 0.15% GBP15

Same trader, same strike rate, same average odds, and a 20.3-fold difference. The arithmetic shows where it goes. The first two outcomes contribute 0.52(0.35) + 0.475(-0.35) = +0.01575. The third (one market in two hundred, losing three times the intended liability) subtracts 0.005(3.00) = 0.015, leaving 0.00075. One rare disaster removes 95% of the edge.

That is the argument for keeping P&L records rather than a win rate.

One practical note if you rebuild this: the outcome probabilities must sum to 100%.

There are a number of important caveats.

Caveats -
1. Treat the output ('s_max') as a ceiling to stay comfortably underneath, not a target to reach.
2. Formula asks whether you will ever hit the drawdown, not whether you will hit it in your next 250 markets.
3. Formula measures drawdown from the starting balance rather than from the running peak, which flatters the result.
4. Formula assumes recent history represents realistic expectations for the current session.
5. If the mean is zero or negative, 's_max' equals 0%: no positive stake is safe without an edge.

Note: LLM used to check the calculations.