Skip to content
citrusquant

Your first strategy

A lemon strategy is one expression that evaluates, every day, to a position matrix: which symbols you hold. The engine turns that into an equity curve. Let’s build one up. Every snippet below runs as-is in the playground against the bundled sample data.

1. A series

The bare name of a panel is a series. close is the daily close for every symbol. On its own it isn’t a strategy — it’s the raw material.

close

2. A signal (a boolean)

Comparisons produce a boolean matrix — true where the condition holds. A classic trend filter: hold a name while its close is above its 20-day simple moving average.

close > sma(close, 20)

sma(of, n) is one of many rolling operators — ema, std, rsi, rolling_max, pct_change, … all take (of, n). See the full op reference.

3. Selecting a few names

Holding everything above its average is a lot of positions. is_largest(of, n) keeps only the n names with the largest value each day:

is_largest(sma(close, 2), 3)

Read it as “hold the 3 names with the highest 2-day average close.” This is the default strategy in the playground — run it, then change 3 to 1 and watch the equity curve concentrate.

4. Combining conditions

Use and / or / not to compose signals. Momentum and a trend filter:

is_largest(pct_change(close, 20), 3) and (close > sma(close, 50))

“Of the names with the strongest 20-day return, hold the top 3 that are also above their 50-day average.”

5. Using fundamentals

The sample dataset also carries a pe panel. Prefer cheaper names by ranking on inverse P/E — rank(of, ascending=...) gives a cross-sectional rank:

is_smallest(pe, 3) and (close > sma(close, 20))

“Hold the 3 lowest-P/E names that are also trending up.”

6. Run it

Paste any of the above into the playground and press Run (or Ctrl/Cmd+Enter). You’ll get an equity curve plus Sharpe, max drawdown, win rate, and more. To understand each number, read Reading a report.

Where to go next