A second strategy under development at Rionda Capital. It applies leverage to the same research that drives GC8. It manages no capital and has no live track record. The simulated results below are shown in full, along with the reasons to discount them.
We are publishing both sizes rather than the flattering one, and the result is not the trade-off you would expect: the larger size returns less and draws down more. Total risk is capped as a multiple of equity, so doubling the size of each position halves how many can be held at once — and in a strategy whose return sits in a small tail of outcomes, holding fewer of them costs more than the extra size adds.
| Over the period | 1% of equity | 2% of equity |
|---|---|---|
| Total return | +8,245.2% | +5,882.7% |
| Annualized return | 460.0% | 391.9% |
| Maximum drawdown | 67.6% | 82.0% |
| Positions | 11,090 | 9,063 |
| Positions closed in profit | 41.9% | 41.2% |
R34 remains in research. The period runs from January 2024 to August 2026, which is as far back as the underlying data reaches, so there is no three or five year record. Returns are calculated from prices actually transacted rather than from a pricing model, across 12,604 positions.
The figures above exclude the cost of trading. They are calculated at prices that transacted, and say nothing about what it costs to cross the spread getting in and out. That cost is not a rounding error here — it is roughly the same size as the edge, so it decides whether the strategy is remarkable or worthless. The table below is the same period at a range of assumptions.
| Cost to enter and exit | 1% of equity | 2% of equity |
|---|---|---|
| None — as shown above | 460.0% | 391.9% |
| 1% round trip | 255.2% | 145.4% |
| 2% round trip | 129.6% | 17.8% |
| 3% round trip | 49.7% | −39.9% |
Annualized return over the full period. Beyond roughly a 3.5% round trip at the smaller size there is nothing left, and at the larger size the threshold is lower still. Establishing where the real number falls is the purpose of the live observation described below.
Returns are also concentrated. The best fifty positions — four in every thousand — average roughly ten times their outlay, while everything else averages close to nothing. Removing those fifty turns the larger size negative. They recur in every quarter of the period rather than arriving in one stretch, but a strategy whose result depends on a tail that thin should be read as one, and losing days are frequent.
The prices are real. The coverage is not complete. Every entry and exit above is a price someone actually paid on the day, taken from transaction data rather than produced by a pricing model. All 19,132 of the underlying strategy's trades were attempted; 12,604 could be expressed at all. The remaining third had nothing tradeable against them, so R34 can only ever carry two thirds of the strategy it is built on.
It is not the strategy as designed. For most of these names, what could actually have been bought had to be held considerably longer than the design calls for — which costs more and delivers less leverage. These are the results of what was available, not of what was specified.
A transacted price is not your fill. Historical bid and ask are not available to us, so each position is valued at a price that traded rather than at the price we would have had to pay to cross the spread at that moment. In this part of the market that difference is material, and it works against the figures above rather than for them.
Two years is all there is. The underlying data does not reach before 2024, so there is no three or five year record and no full cycle to test against. This is why R34 is running as an observer: recording, in live markets, the prices it would actually have transacted at. That comparison is the whole point, and it is not finished.
Historical bid and ask are not available to us, so every figure here is priced off transactions rather than off what we would have been quoted. The cost of crossing that gap is close to the size of the edge itself, which makes it the one number that decides whether R34 is worth running at all.
A price that traded is not the price we would have been given. Real books in this part of the market are thin and move quickly, and the cost of crossing the spread is charged on the way in and again on the way out. A strategy that only works at the midpoint is not a strategy.
Opportunities cluster on volatile days — exactly the days when pricing widens and liquidity thins. Average conditions are not the test. The bad days are the test.
Until all three are answered on a live book we are running ourselves, the figures above stay labeled a simulation rather than a record — and R34 stays in research.
R34 can only be applied to a fraction of the names GC8 trades. Most US equities cannot support the structure this strategy requires, and among those that can, most are not liquid enough to consider.
The structure R34 depends on trades in modest size. Whatever it eventually supports, it will be a small-capacity strategy — and we would rather say so now than discover it with client money.
R34 is under development. It manages no client capital and is not offered. The results above are simulated, not actual: assembled from historical transaction prices rather than produced by trading, on a partial sample and without the cost of crossing the spread. They are to be assumed to overstate what live execution would achieve. Nothing here is an offer to sell or a solicitation to buy any security or interest in any fund.
The research, the assumptions behind it, and what the live observation has shown so far — including where the real fills disagree with the simulation. Under confidentiality, and with the caveats stated plainly.