Renaissance Technologies / 1988 / Quantitative
The exception to the big bet.
Renaissance Technologies and a model of repeated small edges.
Before the outcome was obvious.
Medallion sits awkwardly in a collection of famous trades, and that is why it is useful. The workbook labels the case systematic and gives 1988 as its year, alongside a stylised annual return model. Those inputs do not describe a verified 1988 trade. The conceptual contrast is with a single dramatic prediction: a systematic process attempts to identify, test and repeatedly execute many opportunities. This entry is excluded from the historical comparison chart.
What could have gone wrong
Backtests can overfit; costs, capacity and changing market behaviour can consume an apparent statistical edge.
Repeatability is a different claim from a spectacular outcome, and requires different evidence.
Put the thesis to the test.
Change the assumptions and watch the arithmetic update. These are simplified scenarios; they do not recreate the instrument’s pricing or the path of the trade.
How this model works
Profit ratio = profit ÷ stated position. Exposure ratio = stated position ÷ fund assets. Neither is a measure of maximum loss or a net investor return.
The workbook’s simple annualisation is profit ratio × 365 ÷ duration = 66%. It is linear, not compounded, and is not a repeatable annual return. Fees, financing, collateral, premium carry and changing exposures are not modelled.
What these figures mean.
Source: Legendary Trades Models, Master Comparison!A24:L24; Trade P&L Models, case 21. The following are original workbook inputs, preserved for transparency. They are estimates, not independently audited results.
- Workbook position
- $10bn
- Workbook “gross profit”
- $6.6bn
- Workbook duration
- 365 days
- Workbook fund assets
- $10bn
- Workbook fund return
- 66%
- Workbook position ROI
- 66%
The fund-return field is a separate source claim, not a calculated result of this trade. Its reporting period and gross/net basis are not independently established here. Model ratios are recalculated from inputs rather than copied from rounded source ROI values.