CLAUSE · BACKTEST REPORT · HYPOTHETICAL PERFORMANCE

Half-Kelly momentum

“Half-Kelly Momentum”

Tested over October 2014 to October 2026. On data sealed Oct 15, 2026. Published 15 October 2026.

The contract

1. Universe. Trade only AAPL, MSFT, NVDA, GOOGL, AMZN, META.
2. Signal. Each trading day at 09:35 ET, read each universe member’s 60-day rate of change.
3. Rebalance. Rebalance into the top 3 of the universe by 60-day rate of change, sized by 0.5x Kelly on a 126-day lookback.

Equity vs SPY, dividends reinvested

STRATEGYTHE SAME NAMES, DIVIDENDS REINVESTEDHOLDINGSSPY, DIVIDENDS REINVESTEDlog scale

HOLDINGS3 names at a time, 6 over the run

Tear sheet

TOTAL

+2,002%

CAGR

+28.8%

VOL

26.9%

SHARPE

1.01

SPY SHARPE

0.76

BUY-AND-HOLD SHARPE

1.17

SORTINO

1.45

MAX DD

−37.5%

LONGEST DD

535 days

BETA

1.08

GROSS

90.5%

GROSS MAX

100%

NET

+90.5%

LONG

93.8%

SHORT

0.0%

TRADES

4368

TURNOVER

2,156%

COST DRAG

2.2%

CAPACITY

$11,497,315 (GOOGL)

FINAL VALUE

$2,102,048 from $100,000

Per-clause attribution

What each clause earned while its decision governed the portfolio, holds included; these add up to the total return. The second figure re-runs the backtest with that clause removed and shows how the result would change.

EARNEDWITHOUT

Clause 3

+2,002%−2,002%

About these numbers

Window

Tested from Oct 3, 2014 to Oct 14, 2026. On a copy of the market sealed Oct 15, 2026.

Fills

Orders fill at the next open, 5 basis points each way.

Borrow

Every short pays 0.3% a year on the value of the position for each calendar day it is held. Where our data carries short interest, the rate steps up with short interest as a percent of shares outstanding, using the figure that was public before each session.

An assumption, not a record: we hold no historical borrow rates.

Short interest	Borrow
20% of shares outstanding or more	12% a year
10% of shares outstanding or more	4% a year
5% of shares outstanding or more	1.5% a year
under 5%, or none on file	0.3% a year

Dividends

Cash dividends are credited on the ex-date to whoever held the shares at the close before it: a long receives them, a short pays them, and the cash is reinvested at the next open. Capital gain distributions and dividends paid in stock are not modelled..

Margin

No margin interest, no maintenance calls modelled.

Delistings

A holding that stops trading exits at the last print held.

Halts

Treated like a delisting.

Attribution

Each clause's share of the result is measured by re-running the backtest without it.

Hypothetical performance on historical data. The same contract on the same data and cost model reproduces this number exactly. Backtests are not indicative of future results.