Reading about insider buying signals is one thing. Turning them into a structured, trackable portfolio is another. This guide covers the practical mechanics: when to enter, how much to allocate, when to exit, and how to measure whether the strategy actually works.
Entry rules
Not every CEO purchase is a portfolio entry. You need a filter that balances signal quality with enough frequency to generate meaningful data. Here's what we use:
| Filter | Threshold | Rationale |
|---|---|---|
| Transaction type | Open-market (P) | Voluntary capital at risk |
| Insider role | CEO or CFO | Maximum information advantage |
| Purchase value | > $100K | Material conviction |
| Market cap | > $200M | Sufficient liquidity |
| Multi-factor score | > 65/100 | BUY or STRONG BUY rating |
| Broker availability | Listed on major broker | Executable for retail investors |
After filtering, we typically see 3-8 qualifying signals per week. Not all will be actioned — the multi-factor analysis score determines which pass the final bar.
Position sizing
The simplest approach that works: equal-weight positions. Each qualifying signal gets the same dollar allocation regardless of conviction level.
Why equal weight over conviction-weighted:
- It removes bias. You'll inevitably overweight signals you "feel" good about and underweight unfamiliar ones. Equal weighting forces discipline.
- It's statistically cleaner. When you evaluate performance, equal-weight removes position-size effects from your analysis. A 10% winner and a 10% loser are symmetric.
- It's simpler to manage. One variable (signal count) instead of two (signal count + conviction level).
We use $100 per signal in our tracking portfolio. For a real portfolio, you'd scale this to whatever per-position size makes sense given your total capital — the key is consistency.
A single insider buying signal should never represent more than 5% of your total portfolio. If you have $10,000 to allocate, that's $500 maximum per signal. With 3-8 signals per week, you'll fill a 20-40 position portfolio within a month.
Exit rules
Every position needs a predefined exit. We use a simple asymmetric target/stop-loss system:
STOP LOSS: -15% from entry price → Close position (loss)
// Example
Entry: $50.00
Target: $55.00 (+10%)
Stop: $42.50 (-15%)
// Risk/Reward ratio
Reward: $5.00 per share
Risk: $7.50 per share
Ratio: 0.67:1 → need >60% win rate to profit
Why +10% / -15%?
This asymmetry is intentional. The +10% target is realistic for a 1-3 month holding period on a high-conviction insider signal. The -15% stop-loss is wide enough to avoid being stopped out by normal volatility but tight enough to limit damage from genuinely deteriorating positions.
The math: with +10% wins and -15% losses, you need a 60% win rate to break even. Based on our backtested data across 210 filings, our filtered signals produce a win rate of approximately 65-70%, which generates positive expected value.
Benchmarking: the SPY mirror
Performance in isolation is meaningless. A portfolio returning 8% while the S&P 500 returns 12% is underperformance. You need a benchmark.
Our method: for every signal position, create a matching SPY position on the same day with the same dollar amount.
| Position | Entry | Amount | Exit trigger |
|---|---|---|---|
| ACME Signal | $50.00 on Mar 5 | $100 | +10% target or -15% stop |
| SPY Benchmark | $521.40 on Mar 5 | $100 | Closes when signal closes |
When the signal position hits its target or stop-loss, you close both positions and compare. The difference is your alpha — the excess return attributable to the insider signal rather than general market movement.
This paired benchmarking method eliminates market-timing noise. If the whole market went up 5% during your holding period, your SPY benchmark captures that. Any outperformance above it is genuine signal alpha.
Portfolio management
Diversification
With equal-weight positions and 3-8 new signals per week, you'll naturally build diversification over time. A few rules to enforce it:
- No more than 3 positions in the same sector. If healthcare signals dominate a particular week, cap at 3 and skip the rest.
- No duplicate tickers. If you already hold a position in ACME and a new signal fires, skip it. One position per company.
- Cap total open positions at 30. Beyond this, you're diluting the signal quality with marginal entries.
Rebalancing
There's no periodic rebalancing. Positions enter when signals fire and exit when targets or stops are hit. The portfolio is self-cleaning — winning positions close at +10% and losing positions close at -15%. Capital is recycled into the next qualifying signal.
Record keeping
Track every position with these fields:
Target | Stop Loss | Current Price | P/L % | Status
SPY Entry | SPY Current | SPY P/L % | Alpha
CEO Name | Purchase $ | Score | Opinion
Expected outcomes
Based on published academic research and our own hypothetical backtesting, here are the historical ranges observed. These are not guarantees or projections — actual results may differ materially. Past performance does not guarantee future results:
| Metric | Expected range |
|---|---|
| Win rate | 60-70% |
| Avg. winning trade | +10% (capped by target) |
| Avg. losing trade | -12 to -15% (capped by stop) |
| Avg. holding period | 15-45 days |
| Annual alpha vs S&P | +4 to +8% |
| Max drawdown | -8 to -12% |
These are ranges, not guarantees. Market conditions, signal quality, and sector concentration all affect outcomes. The edge is real but not enormous — this is a supplementary strategy, not a get-rich scheme.
The bottom line
Building a portfolio around insider buying signals is operationally simple: filter, enter, set target and stop, benchmark against SPY, and track everything. The complexity is in the filtering — which is where multi-factor analysis earns its value.
The strategy works because it exploits a persistent informational edge: corporate executives know their businesses better than outside investors, and their personal capital allocation decisions are disclosed by law within 48 hours. The portfolio construction framework simply translates that edge into something measurable.