Fractional Kelly Budgeting for a Long-Only Dynamic Grid Strategy

Summary

This article presents a long-only dynamic grid framework for linear perpetual contracts that separates trade timing from total capital allocation. The grid sets entry and batch exit levels, with spacing constrained by minimum and maximum bounds, recent ATR and estimated transaction costs. A trend filter controls whether new grid positions may be opened, while each filled batch keeps its original exit target. Layer weights decline at deeper grid levels to avoid martingale-like increases in new exposure.

For sizing, the method estimates a fraction of strategy equity by searching historical mark-to-market returns for the highest average log growth, then applies fractional-Kelly and sample-confidence discounts. It includes floating losses in the observations rather than calculating Kelly from completed winning grid trades alone. The article also describes contract sizing, partial-fill accounting, persistent order intent and separate data, risk and safe-halt states. It provides a research and implementation blueprint, not evidence of profitability: the author calls for baseline comparisons and walk-forward out-of-sample evaluation, and notes that drawdowns, sampling error, costs and perpetual-contract risks remain.

Key ideas

• The grid determines when to trade, while the Kelly layer caps the total strategy budget.
• Grid spacing accounts for ATR, cost estimates and explicit minimum and maximum limits.
• The sizing estimate uses mark-to-market equity returns so that unrealized losses enter the sample.
• Fractional-Kelly scaling and a sample-confidence discount reduce exposure when evidence is limited.
• The proposed system requires out-of-sample comparisons and careful order reconciliation before its performance can be assessed.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.