Skip to content

Every trade is decided by a rule you can read.

Castle Capital runs rules-based strategies on your own Alpaca brokerage account. A strategy is a decision tree written out in full: conditions in, target allocation out. It is evaluated once per trading day, and the resulting orders are placed through Alpaca. Nobody here decides what your account buys.

What a rule looks like

A branch from a strategy file on this platform, and the closed set of signals any rule may reference.

A defensive branch, as written

Read top to bottom; the first condition that passes wins
(weight-equal
 [(if (> (rsi "SPY" {:window 10}) 80)
    [(asset "TLT")]
    [(if (> (max-drawdown "QQQ" {:window 60}) 10)
       [(asset "BIL")]
       [(asset "QQQ")])])])
First condition If SPY's 10-day relative strength is above 80, hold long-duration Treasuries.
Otherwise If QQQ has fallen more than 10% from its 60-day peak, hold Treasury bills.
Failing both Stay invested in QQQ.

Every number above is a parameter someone chose and can be pointed at: the windows, the thresholds, the tickers. There is no model weight to interpret and no state carried between days. The same prices on the same date always produce the same portfolio.

What a rule may look at

Six indicators, nothing else
Relative strength over a window rsi
Moving average of price, simple or exponential moving-average-price
Moving average of daily return moving-average-return
Total return over a window cumulative-return
Peak-to-trough decline max-drawdown
Latest traded price current-price

A rule cannot reference anything outside this list. There is no news feed, no sentiment model and no manual input. If a signal is not above, no strategy on this platform can act on it.

The decision path

What happens between a market close and an order reaching your account.

01The rule file

Produces A named strategy

Each strategy is written out as a decision tree: the conditions, the branches, and the target allocation at the end of every branch. The same inputs always produce the same portfolio, and the file is version controlled, so a change to a strategy is a change someone can point at.

02Market data

Guards against Stale or partial data

Before anything is evaluated, daily price history is fetched and cached for every symbol the rules reference. If a series is missing or stale the run stops. The platform does not fill a gap with an estimate and trade on it.

03Evaluation

Runs Once per trading day

Conditions are tested in the order they are written: a relative-strength reading, a moving-average comparison, a drawdown threshold. The first branch that passes determines the target weights. Nothing is scored, blended or overridden after the fact.

04Sizing

Capped by Your plan's limit

Target weights become dollar amounts against your account equity, capped by the capital limit on your plan. Depending on the strategy, weights are equal across the basket, explicitly specified, or set inverse to each holding's recent volatility.

05Execution

Venue Alpaca Securities LLC

Only the difference between the target and what you already hold is traded. Orders are submitted to Alpaca as notional market orders, and a run gate records the day so the same strategy cannot execute twice.

The five strategies

Each one answers a different market condition. Each one has a cost when that condition is absent.

Momentum

Answers Sustained uptrends

Holds the index and sector funds that are already trending and stays with them while the trend holds. It gives back part of a move at every turning point, which is the price of not trying to predict them.

Typically holds Equity index ETFs

Mean reversion

Answers Range-bound markets

Buys after a sharp move away from a recent average and exits as the gap closes. It suits a choppy market and is the strategy most exposed to a move that keeps going instead of turning.

Typically holds Index and sector ETFs

Volatility

Answers Elevated volatility

Takes positions in volatility-linked funds when relative-strength and drawdown readings say the market is under stress. Those funds hold futures and lose value while conditions are calm, so the rules hold them only while the trigger persists.

Typically holds Volatility-linked ETFs

Tactical rotation

Answers Regime changes

Moves between sectors, regions and asset classes on relative-strength comparisons rather than a house view. The objective is to be out of what is weakening, not to call the next leader early.

Typically holds Sector and regional ETFs

Capital preservation

Answers Falling markets

When drawdown and trend conditions trigger together, the target allocation moves into short-duration Treasury and cash-equivalent funds. Holding cash equivalents is a position, and this is the branch that takes it.

Typically holds Treasury and cash-equivalent ETFs

Strategies are evaluated independently and rebalance daily. Several use leveraged or inverse ETFs: that leverage sits inside the fund, not as margin in your account, and those funds can lose value quickly and are not intended to be held through every market. Nothing on this page is a recommendation, and past performance does not guarantee future results.

Controls and limits

The constraints that sit around every run, whatever the strategy decides.

Capital limit

Your plan sets the maximum capital the system will deploy. The rebalancer sizes to that figure, not to your whole balance, so the rest of the account is never in play.

Trading-calendar gate

Every run is checked against the exchange calendar, including holidays and early closes. Outside a valid session nothing is submitted.

Duplicate-run protection

Each completed run is recorded. A second trigger for the same strategy on the same trading day is refused rather than doubling a position.

Your keys, your account

Trades settle in your own Alpaca account and the cash never passes through Castle Capital. Your API credentials are encrypted at rest, and revoking them at Alpaca stops the platform immediately.

Backtested before it is offered

Every strategy is run over historical prices before it reaches the platform, and you can run the same backtest yourself once you are signed in. A backtest is a simulation of a rule against past data, not a forecast.

Evaluation Python service
Price data Daily bars, cached
Portfolio snapshot 4:00 PM ET
Backups Hourly and nightly

What this platform does not do

Worth reading before the parts that sound good.

It does not give advice

Castle Capital LLC is not a registered broker-dealer or investment adviser. You choose the strategy; the platform executes the rules attached to it. It does not assess whether that strategy suits your circumstances.

Nobody trades on your behalf

There is no discretionary desk and no manual override. If no condition fires, the platform does nothing that day, including on days when doing something would have been better.

No options, futures or margin

Strategies allocate across listed ETFs only. Where leverage appears it is inside the fund itself; the platform does not borrow in your account.

It is not a speed advantage

Decisions are daily and orders are ordinary market orders. This platform does not compete on latency and makes no claim to a sub-second edge.

It does not remove risk

Rules remove hesitation, not loss. Every strategy here can lose money, including the ones designed to be defensive, and a rule that worked for a decade can stop working without notice.

Run these rules on your own account

Connect an Alpaca account, choose a strategy, and set the capital you want the rules to manage. You can stop it at any time from your account settings.