R-01Foundational
Strategy Research
Trading ideas become precise, rule-based hypotheses — defined well enough that a machine could execute them and a skeptic could falsify them.
Lone Summit Trading Company is an independent quantitative research desk for CME futures. We stress-test systematic strategies against the harshest questions we can ask — and keep only what the evidence supports. Most ideas don't make it through. That's the point.
A promising backtest is a hypothesis, not a result. Before any capital is ever placed behind an idea, it must pass through a sequence of increasingly unforgiving tests.
A hypothesis about how markets behave — stated precisely enough that it could be proven wrong.
Entries, exits, sizing, and risk limits become unambiguous logic that a machine can execute without interpretation.
Years of clean market data, with transaction costs, slippage, and contract rolls modeled rather than ignored.
The strategy is replayed against history to produce its first piece of evidence — and only that.
Parameter spaces are mapped in full. Broad performance plateaus matter; knife-edge peaks do not.
Parameters, sessions, instruments, and costs are systematically perturbed to find exactly where the idea breaks.
Everything is re-run on data the strategy has never touched. The only test that truly counts.
Thousands of re-orderings and simulations expose the full drawdown distribution — not one flattering path.
The system runs forward against live market conditions with no capital at risk — and no way to fool itself.
If an idea makes it this far: minimal size, full instrumentation, and expectations logged in advance.
Markets change regimes, edges decay, and assumptions expire. Validation never formally ends.
Six disciplines, each one built to attack an idea from a different direction.
Trading ideas become precise, rule-based hypotheses — defined well enough that a machine could execute them and a skeptic could falsify them.
History is replayed with transaction costs, slippage, and contract rolls accounted for — not the way brokers draw it.
Large parameter spaces are searched systematically, with performance mapped across the whole surface — not cherry-picked at its best point.
Parameters, sessions, instruments, and costs are deliberately perturbed to find where an idea breaks.
Trade order, timing, and fills are re-randomized to expose the drawdown distribution a single run would hide.
Bull, bear, chop, calm, crisis — performance is studied condition by condition to learn where an edge lives, and where it doesn't.
A research memo should read like evidence. Four representative sections are shown below — every figure illustrative, every limitation stated.
The operating boundary a study is read against — account rules, cost assumptions, and market access. Every later section is interpreted in this context.
| Parameter | Studied value |
|---|---|
| Daily loss limit | $2,500 |
| Trailing drawdown | $3,000 |
| Maximum allocation | 3 mini / 6 micro |
| Markets studied | ES, NQ, GC, CL |
| Trading windows | US open + London |
Risk-adjusted measures are shown together so no single metric carries the conclusion.
Trade-order reshuffling estimates how historical sequence variation could change the drawdown profile. It does not model every live-market risk.
Each configuration is documented individually, then reviewed as part of the whole portfolio.
| Market | Window | Allocation | Role |
|---|---|---|---|
| ES | US open | 1 mini | Core |
| NQ | US open | 2 micro | Diversifier |
| GC | London | 1 mini | Non-equity |
| CL | US open | 2 micro | Non-equity |
Optimization can discover good parameters. Research determines whether those parameters mean anything. The failure modes below end more trading systems than bad ideas do — so they get their own section.
A model tuned until it fits the noise fits the past perfectly — and predicts nothing.
Every extra dial makes the backtest more flattering and less true.
Test five hundred ideas, and the single best one will look brilliant by luck alone.
A handful of trades is an anecdote. Evidence requires enough observations to mean something.
An edge that lives in one market environment may simply not exist in the next one.
Nothing eliminates overfitting. Finding the flaw is the research.
Lone Summit builds its own research software — data pipeline, search engine, and analysis stack — because the questions worth asking don't fit inside retail charting tools.
Internal & proprietary. Not licensed, not exposed.
A single validated strategy is one data point. The more interesting question is how several independent systems behave together — especially when conditions turn hostile.
The research is published so the method can be judged. The moment findings became a product, the incentive to flatter them would win.
Lone Summit Trading Company is an independent quantitative research company focused on systematic futures trading, trading technology, and proprietary research infrastructure. Work is concentrated in liquid CME futures — equity index futures, metals, and other deeply liquid contracts — and every idea must earn its way through validation before any capital follows.
The objective is not a perfect backtest. It is conclusions that remain understandable, defensible, and clearly bounded. Research first — validate everything.