Midas Touch TechnologiesMTT

Stop wondering if it would’ve worked.

Describe any trading strategy — or paste one you found — and our agent backtests it against decades of real data, compares it to what you’re doing now, and trades it automatically under your rules.

evidence first. execution only when it earns it.

One conversation, from idea to evidence.

A real session: describe a strategy in one sentence, the agent writes it up as a testable configuration, runs the backtest, and lays the evidence next to the baseline — equity curve, every trade, every signal.

captured from a live session, sped up · backtested results are historical and hypothetical, not live returns

01A testable configuration

One sentence in, and the idea comes back written down: symbols, timeframe, the exact entry and exit conditions, position sizing, starting capital. Nothing is inferred behind your back, and a validation pass confirms every reference resolves before a single bar is read.

The saved aapl_sma_crossover_2 strategy: symbols AAPL, 1d timeframe, entry and exit expressed as SMA-crossover conditions, 100% of capital per position, 100,000 USD capital, and an analysis panel reporting that syntax is OK and all four data references resolve

02Every trade, on the curve

Then the engine runs it against decades of daily history — this one goes back to 1984. All 27 trades are marked where they happened and every holding period is shaded, so you can see when the strategy was actually in the market rather than trusting a summary number.

The backtest equity curve for the AAPL SMA crossover strategy from September 1984 to August 2026, with entry and exit markers on each of its 27 trades and holding periods shaded

03Scored against a baseline

And every run is compared to what you'd have had by doing nothing. This one lost: holding Apple over the same forty years returned more, with a shallower drawdown. The agent leads with that instead of burying it.

Backtest results comparing the AAPL Golden Cross strategy against AAPL buy and hold across return, CAGR, win rate, max drawdown, Sharpe ratio, and market exposure, with a written takeaway explaining that the strategy underperformed buy and hold

Sometimes the evidence says stick with the index.

We’ll tell you that too. No trade happens without a backtest, and no trade happens outside rules you set.

live today

Strategy authoring in plain English, backtesting against decades of market data, and full evidence review — charts, trades, and every signal inspectable.

in early access

Portfolios that paper-trade their strategies automatically, marked to market daily.

next

Brokerage execution — always under thresholds the user sets, never without a backtest.

From idea to execution.

Describe it.

In plain English, or paste a link. No code.

See the evidence.

Backtests on real data. Risk-adjusted comparison to your baseline.

Let it run.

Automatic execution and daily rebalancing — under thresholds you set.

Want the details?

Full metrics, distributions, projections, parameter sweeps. Every number inspectable, nothing hidden behind a score.

Just want the answer?

Does it beat what you're doing now, risk-adjusted? One comparison, in plain terms, against your current default.

paste the video. get the backtest.

Ideas come from anywhere.

A video that seemed smart, a thread you half-remember, a hunch about a sector. Paste the link — the agent turns it into a real, testable strategy and runs it against history like any other.

found this — is it actually any good?youtube.com/watch?v=golden-cross
Built it. 214 trades over ten years — here’s the evidence.

on the roadmap · not yet available

The agent will get better the longer you use it.

Today every conversation starts fresh. We’re building a scratchpad: a running record of what you’ve tried and how it turned out. A strategy that didn’t clear your baseline is still evidence — what it tells you about entries, exits, and when the whole approach works rolls forward into the next strategy you try.

You test a trend-following entry, then five variations of it — a tighter trigger, a confirmation filter, a shifted lookback.
next time

Weeks later you start tuning a different trend strategy the same way. The agent says so up front: last time, every tuned version finished behind the one you left alone.

A breakout strategy comes back winning 38% of its trades, and you drop it on the spot.
next time

The next low-win-rate result arrives with the payoff ratio leading instead of the win rate — because for trend following, losers cut early and winners left to run is the design, not the defect.

A defensive rotation looks mediocre across the full backtest, so it never makes the cut.
next time

Months on, you ask for something that holds up in a crash. The agent brings that one back — flat for years, and the only rule you've tested that gained while the index fell.

Who we are.

Midas Touch Technologies is a two-person, bootstrapped software company founded in 2026. Everything on this page — the agent, the backtest engine, the data pipeline under them — is one product, built and run by us.

Connor Smith

Founder

Seven years building data platforms and LLM products — most recently a fintech's company-wide AI analytics assistant, a multi-agent system over a governed semantic layer. M.S. Data Science, Rice. Builds the agent, backtest engine, and warehouse here.

Sacha Le Clainff Alonzo

Founding Engineer

M.S. Data Science, Rice, with a computer science background. Has built full-stack AI applications and quantitative trading systems. Owns the control plane here: schema, auth, identity, and the local/CI database workflow.

More about the company →

Ready to stop wondering?

Join the waitlist