How it works.

One engine runs every strategy. Nine tests decide which strategies join.

The engine, trade by trade

One backtest trade, as the engine logged it and as it looked on the chart.

engine.logmodel account, backtest replay, 1 Mar to 3 Mar 2026, New York time

1 Mar 22:00bar2-hour bar closed # the strategy reads its own chart
1 Mar 22:00signallong # buy stop at 44.69, model portfolio at $100,000
1 Mar 22:00size2 micro contracts # inside the risk budget
2 Mar 00:00orderentry STOP filled 44.69
2 Mar 00:00orderstop STOP working 40.14 # placed on fill
2 Mar 00:00ordertarget LIMIT working 46.88 # placed on fill
2 Mar 00:00ledgerposition open matches model account
each barorderstop and target re-set # to 39.77 / 47.09 by the exit
3 Mar 04:00filltarget LIMIT filled 47.09 # target reached
3 Mar 04:00orderstop cancelled # one-cancels-other
3 Mar 04:00ledgerflat matches model account

A bar’s time is its close. Picked to show how an order works, not a typical result. Hypothetical*

The same crude oil trade on its 2-hour chart. Hypothetical*

The path of every trade

If you switch copying on, each position is copied into your account as it happens, at your trading size, and closed when the model closes it.

Nine tests before a strategy joins

We publish how each strategy is tested, formula by formula. A strategy is judged on years its rules were not fitted to, and its result is marked down for every version tried before it.

Confidence grows with time
$t = SR \cdot \sqrt{T}$
The same edge over more time, or more independent trades, is harder to explain by luck.
1The entry is tested alone
$E_{N} = \frac{\sum_{i} \frac{{MFE}_{i, N}}{{ATR}_{i}}}{\sum_{i} \frac{{MAE}_{i, N}}{{ATR}_{i}}}$
Before any exit exists: over the N bars after each signal, the largest move in the trade’s favour against the largest move against it, both in units of ATR (average true range).
2Judged on years it was not fitted to

in-sampleout-of-sample 1out-of-sample 2
$WFE = \frac{1}{K} \sum_{k = 1}^{K} \frac{R_{OOS, k}}{R_{IS, k}}$
Fitted on the first window, then judged unchanged on K later windows: return out of sample as a share of return in sample.
3It has to work on related markets
$\Pr [pass ≥ k of m ∣ no edge] = \sum_{i = k}^{m} (\binom{m}{i}) p^{i} {(1 - p)}^{m - i}$
A rule with no edge seldom passes on several markets at once. Related markets move together, so the formula understates luck; the test is read that way.
4Small changes to its settings
$ρ = \frac{\underset{θ^{'} \in N (θ)}{median} S (θ^{'})}{S (θ)}$
a peaka plateau

Every setting nudged up and down. Nearby settings should score almost as well: a plateau, not a peak.
5Thousands of unlucky versions of its past
$D_{95} = Q_{0.95} [\max_{t} (1 - \frac{E_{t}^{(b)}}{\max_{s \leq t} E_{s}^{(b)}})]$
Trades reshuffled and a share removed at random, over thousands of runs (b). We plan around the bad tail of that distribution, not the backtest.
6Thousands tried, luck priced in
$DSR = Φ [\frac{(\hat{SR} - {SR}_{0}) \sqrt{T - 1}}{\sqrt{1 - {\hat{γ}}_{3} \hat{SR} + \frac{{\hat{γ}}_{4} - 1}{4} {\hat{SR}}^{2}}}]$
N tried1001,0002,000DSR99.7%97.2%95.6%

The result must beat the best that N tries with no edge would show, skew and fat tails included.
7The bad tail, measured directly

${ES}_{0.95} = E [r_{t} ∣ r_{t} \leq q_{0.05}] = - 4.3 %$ ${\hat{γ}}_{3} = + 0.63$

The average of the worst 5 of 101 months, not a single number at a cut-off. A positive skew: the largest months were gains.
8Not the same bet twice
${\hat{SR}}_{c} > {\hat{ρ}}_{c, P} \cdot {\hat{SR}}_{P} {\hat{ρ}}_{c, P} = corr (r_{c}, r_{P})$
A candidate joins only if its Sharpe ratio exceeds its correlation with the portfolio times the portfolio’s own: the condition under which a small allocation to it raises the portfolio’s Sharpe ratio.
9Sized to its worst case, then sized together
$f_{B} = \max {f : D_{95} (f) \leq B}$
Each strategy gets the largest size f whose 95th-percentile drawdown stays inside a fixed budget B. Then the whole portfolio is tested the same way, all strategies at once.

The pipeline in nine formulas. Its figures come from the simulation. Hypothetical*

How few pass: an illustration

Each dot is one candidate.

Illustration, not a count of a real pipeline run.

Each test’s pass mark is written down before a strategy is tested, so the bar cannot move after a result is seen.

Who we are

Eon Quant builds and runs the engine behind its model portfolios, and the research that decides which strategies join them.

The name came from a conversation at the very start. Banks and professional hedge funds see market data the moment it is released, through servers placed beside the exchange. Their machines have finished trading “eons” before the same data reaches a smaller firm. Eon Quant is named after that gap, as a nod to the small player. Rather than enter a crowded, uphill race it would lose, it works on longer horizons, with trades that last from hours to weeks. It picks a game that favours patience: rules-based strategies built to hold up through market cycles, careful risk management, and time.

What we never do

Hold subscribers’ money.
Lock anyone in.
Promise a return.
Trade a subscriber’s account ourselves.
Give personal advice.
Call a backtest a track record.
Leave out a losing month.

Questions

Other questions are best taken on the introduction call.

Why these markets?

Equity, futures and forex markets are deep and allow short positions as well as long; futures and forex trade for most of the day. Futures are exchange-traded and centrally cleared: every participant holds the same standardised contract.

What if the engine fails mid-trade?

It saves its state, so after a restart it resumes managing the model portfolio’s open positions. Anything it cannot reconcile stops new trades from being published and alerts our team.