Lodestar Autonomous Multi Market EA

Eleven positions across seven markets. One risk dial. Volatility targeting that was measured, not promised.

Measured 2020-2026: $100,000 became $186,011 - +9.93% a year, Sharpe 0.58, worst drawdown −31.2%, one losing year in seven.
It trades: US500, Nasdaq, DAX, Nikkei, Gold, Silver, Crude Oil, plus a permanent four-leg short-dollar hedge.
Minimum capital: $25,000.
Forecasting: none. The market-timing signal ships switched OFF, and further down you will find out why.

What it did when the market actually broke

A backtest average tells you nothing about the days that decide whether you stay invested. So here is Lodestar against simply buying and holding the same seven markets, over the same period, with the same broker's data and the same financing charged to both sides.

In the crash that mattered, it lost a quarter of what the market lost - and 2020 still finished +22.3%. In the 2022 bear market it fell about the same as everything else; volatility targeting is not a forecast and does not dodge a slow grind. But both books peaked in November 2021, and Lodestar made a new high in March 2024 while buy & hold waited until June 2025. Fifteen months earlier, out of the same hole.

No model is involved on either side of that table. One column is the binary's own tester equity curve; the other is seven daily closes carrying the same overnight financing, charged the same pessimistic way.

The risk dial

Most portfolio EAs use fixed lots or a flat percentage of equity, so your real risk drifts with the market instead of staying where you put it. Lodestar re-sizes every position daily from 60-day measured volatility. The dial is the product.

Read the three rows as one sentence: ask for a risk level and you get it. The 30 setting does earn more money - and it charges you a 42.6% drawdown and a lower Sharpe for the privilege. The default sits at 20 because that is where the risk-adjusted return peaks, not because it is the biggest number.

Every year, including the bad one

Strategy Tester, 2020.01.01 - 2026.07.24, $100,000, every tick, no preset file. This is what happens when you drag it onto a chart and change nothing.

2020 +22.3%  |  2021 +7.7%  |  2022 −21.4%  |  2023 +13.7%  |  2024 +26.3%  |  2025 +20.4%  |  2026 +4.0% (to 23 July)

The account reached $146,949 by November 2021, gave a lot of it back through 2022, and made a new high in March 2024. From that new high to the end of the test it added another +25.9%; 2024 and 2025 together were +52.1%.

Two things you should size up before buying, because we would rather you heard them from us:

1. Twenty-eight months is a long time. That is how long the account sat below its November 2021 peak. Buy & hold sat there for 43, so this is the shorter road, not a comfortable one. If a year and a half of a flat-to-falling equity curve would make you switch the robot off at the bottom, the 10 setting exists for exactly that reason - it cut the worst drawdown to 14.1%.

2. The start date matters more than the average. Across all 1,436 twelve-month holding periods inside the test: worst −27.9%, median +13.8%, best +44.3%, and 30% of start dates were still under water a year later. At a 20% volatility target, twelve months is simply not long enough for an average to show up. This is a multi-year instrument.

The test peaked at $201,755 and finishes at $186,011, because the window happens to end inside a dip. We publish the closing figure rather than the peak.

Financing: the cost you can actually do something about

Overnight financing is the largest single deduction in any multi-asset portfolio strategy, and almost nobody shows it to you. Lodestar puts your real holding cost on the chart, live, as a percentage of equity per year.

Broken down, each index swap turns out to be the published central-bank rate plus a flat 4.4% a year of house margin - charged on the long side and the short side. The policy rate you cannot change. The 4.4% you can, by choosing a different broker, and on this book every point of margin you shave back is worth roughly two points of annual return. Competitive desks quote base + 2%.

Two more notes in your favour. Prefer CASH index CFDs to futures CFDs, because cash contracts pay you the dividend adjustment on the long index legs. And MetaTrader's tester charges today's financing rate against every past year, including the years when rates were near zero - so the real historical cost was lower than everything on this page assumes.

Why the $25,000 minimum is real

Eleven positions need eleven minimum lots. Gold costs roughly $4,000 of exposure per minimum lot, oil about $4,500 - so on a small account the expensive markets simply cannot open, and the ones that drop out are not the ones you would choose.

The last column is the whole point. At $2,000 the four currency positions that acted as the crash cushion never open at all - and that hedge is the component which held up in every robustness test we ran. Below the floor the outcome depends on where your equity happens to land relative to each market's lot grid, which makes it unpredictable rather than merely smaller. If your account is under $25,000, this is not the right product for you yet.
Note: Returns scale smoothly from +7.25% at $25,000 up to +9.93% at $100,000 as position-sizing granularity improves with higher capital.

What Lodestar does not do

It does not predict direction. A market-timing signal is built in and ships switched OFF, because it did not beat scrambled history: on block-shuffled prices - which keep the drift and the volatility but destroy all predictability - the timing rule scored higher on the noise than on the real data. We left the code in the product so you can switch it on and check that for yourself.

No martingale, no grid, no averaging down, no hidden recovery mode. Nothing doubles after a loss. Position size comes from measured volatility and nothing else.

It is not a defensive product. It promises a risk level, and the first table is the evidence that it delivers one.

The console, and the brakes behind it

Live financing tracker - your real annualised swap drag, in % of equity, on the chart.
A protective stop on every position, roughly six average daily ranges away, resting on your broker's server so it works while your computer is off. It is a gap brake for accidents, not the risk control - risk is controlled by SIZE.
Daily loss brake at 12% - closes everything and stays flat until the next daily bar.
Total drawdown halt at 50% - flat and stopped until a human re-attaches it. Automatic restarts after a 50% loss are how people lose the rest.
It does not fight you. Close one of its trades by hand and it leaves that market alone until the next rebalance.
Restart-safe, keyed to your account number. After a reboot you do nothing.
A twelve-page manual built into the console, plus a help panel on every card.

Requirements

MetaTrader 5, hedging or netting. Attach to ONE chart only - for example US500 on D1 - and it runs all eleven positions from there. Do not run several copies.
Symbols: US500, USTEC, DE40, JP225, XAUUSD, XAGUSD, XTIUSD, AUDUSD, NZDUSD, EURUSD, GBPUSD. Your broker's spelling is detected automatically (US500 = SPX500 = US500.cash and so on), and any market your broker does not carry is skipped with its risk shared out to the rest. Short selling must be permitted on the four FX pairs.
Timezone: nothing to set. It works on daily bars and reports your broker's own clock on startup.
Capital: $25,000 minimum. No VPS is required, but the terminal has to be running for the daily rebalance.

Risk warning.
This is a tool, not investment advice.