How the Bots Work
Swing Trader is the most active of the four AI portfolios. It holds positions for 1–10 days, actively managing a diversified book of equities, ETFs, and options. It reads the market throughout the day and adjusts accordingly — it's built for the trader who wants to see active, responsive decision-making rather than a buy-and-hold approach.
Every run follows a three-step process:
- Assess the regime — Classify current market conditions as BULL, BEAR, NEUTRAL, or CRISIS based on options flow, dark pool activity, sector momentum, and macro signals.
- Review the portfolio — Evaluate each open position against the current regime and decide whether to hold, reduce, or exit.
- Scan for new setups — Identify new opportunities that fit the current regime and sizing rules.
Every decision is fully documented with the rationale behind it. The AI must justify every action against the current regime — there are no unexplained moves.
Swing Trader is the most flexible of the four bots. It can trade:
- Individual equities
- Sector ETFs
- Inverse ETFs (for bearish exposure)
- Long calls and long puts
- Call spreads and put spreads
The instrument chosen for each trade depends on the setup — a high-conviction directional play might use a long call, while a broader market hedge might use an inverse ETF.
The regime is Swing Trader's read on the overall market environment. It directly controls how much of the $100,000 portfolio is deployed at any given time:
| Regime | What it means | Capital deployed |
|---|---|---|
| BULL | Strong upside momentum, conditions favor long exposure | Highest |
| NEUTRAL | Mixed signals, no clear directional edge | Moderate |
| BEAR | Deteriorating conditions, risk-off posture | Lower |
| CRISIS | Severe market stress, capital preservation mode | Lowest |
In CRISIS mode the bot retreats heavily to cash. This is intentional — protecting capital in a drawdown is prioritized over chasing trades.
Within the overall capital deployment range, positions are sized based on conviction. A setup where multiple signals align strongly gets a larger allocation than a speculative idea with only partial confirmation. No single position is allowed to dominate the portfolio — diversification is enforced as part of the sizing rules.
- Real-time options flow and sweep alerts — large institutional orders that signal where smart money is positioned
- Dark pool prints — off-exchange block trades that reveal accumulation or distribution by major institutions
- Gamma exposure levels — where market makers are hedged and where price tends to pin or accelerate
- Sector ETF performance and rotation signals — which sectors are seeing inflows vs. outflows
- Macro indicators — volatility index (VIX), bond yields, and crude oil prices
- Individual equity fundamentals — price, volume, and technical levels
- Pre-market and after-hours price action
Swing Trader is active throughout the trading day, re-evaluating the portfolio as conditions change. It's the most active of the bots by design — built to react rather than set-and-forget. The exact schedule is proprietary.
Yes. The bot is engineered to learn from its own track record — analyzing closed positions, identifying patterns across trades, and refining its judgment over time. It actively applies what it has learned to future decisions.
The lessons themselves are proprietary and stay private, but they shape every decision and every piece of reasoning you see the bot make on its page.
Value Investor doesn't trade — it invests. It scans for businesses with durable competitive advantages, strong free cash flow, and undervalued balance sheets. It ignores short-term volatility, buys into fear, and holds with conviction for weeks to months. It's designed for patient capital that doesn't need to check prices every hour — the philosophical opposite of Swing Trader.
Every candidate is evaluated against a strict value checklist:
- Is there a meaningful margin of safety between the current price and intrinsic value?
- Is the business generating strong free cash flow?
- Is the balance sheet healthy — manageable debt, adequate liquidity?
- Is management allocating capital well — reinvesting wisely, not destroying value?
Only businesses that pass all criteria are considered. The AI would rather miss an opportunity than overpay — capital preservation is a core principle.
Equities only. No options, no leverage, no inverse positions. Long-only. This keeps the strategy true to value investing principles — it only owns businesses it believes are worth more than they cost today.
Value Investor uses a different regime framework than Swing Trader — instead of market momentum, it assesses the investment environment for long-term ownership:
| Regime | What it means |
|---|---|
| ACCUMULATE | Valuations are attractive, macro supports ownership, margin of safety is sufficient — actively deploying capital |
| HOLD | Existing positions remain sound but conditions don't justify adding new exposure |
| REDUCE | Valuations stretched or macro risks elevated — trimming positions |
| AVOID | Not the time to own equities — retreating to cash until conditions improve |
- Discounted cash flow (DCF) valuations — intrinsic value estimates based on projected free cash flow
- Key financial ratios — P/E, P/B, debt-to-equity, current ratio, return on equity
- Free cash flow yield — how much cash the business generates relative to its market price
- Profit margins and earnings growth trends
- Revenue and earnings history across multiple years
- Balance sheet strength metrics
- Analyst estimate ranges for future earnings
Notably absent: options flow, dark pools, technical levels. Value Investor deliberately ignores short-term market noise and focuses exclusively on fundamental business quality.
Value Investor runs on a much lighter cadence than Swing Trader. This is consistent with a strategy that isn't reacting to every market tick. Patient capital doesn't need a hyperactive schedule.
Yes — the same self-improvement system as Swing Trader, but tuned for value investing. The focus here is on whether the original thesis was correct, whether the entry valuation was truly attractive, and whether the holding period was appropriate. Over time it builds a record of which fundamental signals led to good outcomes and which ones were misleading.
Earnings Trader is a specialized event-driven strategy. It studies upcoming earnings reports, tries to predict whether the market reaction will be positive or negative, and enters positions ahead of the report. It makes a hold-or-exit decision the next morning and closes everything by end of day. No overnight holds past the earnings event.
It also plays the run-up into earnings: in some setups it enters earlier to capture rising implied volatility and exits before the report is released.
It only operates around earnings — when there are no upcoming reports in scope, it sits in cash.
The main play trades through the report in three phases:
- Pre-close entry (day of earnings) — Screen upcoming reports, score each setup, enter positions in the highest-conviction names before the close.
- Morning hold-or-exit (next day) — Assess the actual earnings result, how much implied volatility has collapsed, whether the move is playing out as expected, and decide whether to take profits immediately or hold through the day.
- Pre-close exit (following day) — Everything still open gets closed. No exceptions.
There is also a second play: the pre-earnings run-up. Implied volatility often expands as a report approaches. When the bot sees a setup with a good chance of gaining value ahead of the event, it can enter earlier and exit before the report is released — so it takes no exposure to the earnings result itself.
Earnings Trader has the widest options toolkit of the four bots:
- Long calls and long puts
- Call spreads and put spreads
- Straddles (when direction is uncertain but a big move is expected)
- Iron condors (when the market is pricing in a move that won't materialize)
- Equities when options are prohibitively expensive
All positions are defined-risk — no naked short options. The maximum loss on any trade is known before entry.
Earnings Trader uses its own regime labels that reflect the overall quality of the earnings setup environment:
| Regime | What it means | Capital deployed |
|---|---|---|
| STRONG SETUP | Multiple high-conviction earnings setups available with aligned signals | Highest |
| MODERATE SETUP | Some good setups but signals are mixed or conviction is lower | Moderate |
| WEAK SETUP | Setups available but conditions don't favor aggressive positioning | Minimal |
| AVOID | No compelling setups or macro conditions unfavorable | Stays in cash |
Each candidate is scored across multiple dimensions, with setups selected only when several signals align together. The model weighs:
- Earnings history — the company's track record around prior reports
- Smart-money positioning — what institutional flow and dark pool data are signaling ahead of the report
- Options market signals — including how richly the upcoming move is being priced
- Technical and momentum context — where the stock sits going into the print (a stock that has already run up into earnings faces a higher bar than one that's been beaten down)
The exact weighting and combination logic is proprietary, but the principle is simple: no single signal triggers a trade — alignment does.
The morning decision uses a strict framework. The key factors are IV crush (how much implied volatility has collapsed after the report), the size of the actual move versus the implied move, and whether price action is confirming the thesis.
If IV has collapsed enough to neutralize the trade's edge, if the actual move was meaningfully smaller than what was priced in, or if price is rejecting at key levels — profits are taken immediately.
If the earnings result was genuinely better or worse than expected, price action is constructive with volume confirming the move, and remaining directional potential outweighs the IV decay — the position is held through the day.
Specific thresholds and weightings are proprietary, but the discipline is firm: the morning decision is rule-driven, not narrative-driven.
Before an earnings report, options are expensive because nobody knows how large the move will be — traders pay up for that uncertainty. This elevated uncertainty is reflected in high implied volatility (IV).
The moment the report is released, the uncertainty resolves. IV collapses almost instantly — this is called IV crush. Even if the stock moves in the right direction, the drop in IV can erode or eliminate the value of the options position. A call buyer can be right about direction and still lose money if IV crushes harder than the stock moves.
Earnings Trader monitors IV rank before entry to avoid paying too much for options that are already pricing in extreme moves, and uses the morning IV reading to inform the hold-or-exit decision.
Earnings Trader only runs around earnings events — ahead of a report to evaluate setups, and afterward to manage and review positions. The exact schedule is proprietary.
Yes — and the self-improvement system here is especially detailed because Earnings Trader makes more discrete decisions per trade than the other bots. Over time it builds a pattern library of which setups tend to fade at the open versus run all day, and which IV environments make options too expensive to trade profitably.
Momentum Quant is a systematic, rules-based strategy inspired by how professional quantitative funds identify and ride trending stocks. Unlike the other bots — which use narrative reasoning to interpret market conditions — Momentum Quant applies a strict mechanical rulebook. There's no discretion, just criteria.
The philosophy is simple: buy strength, cut weakness fast, and let winners run. Price is the signal that matters.
Every decision runs through a fixed checklist. To qualify for entry, a stock must:
- Be in a confirmed uptrend
- Outperform the S&P 500 in recent weeks
- Have no earnings report coming up soon
The bot screens hundreds of stocks daily, keeps only those passing all three gates, ranks the survivors by relative strength, and selects from the top of the list. No story, no interpretation — just whether the name clears the bar.
Momentum Quant is the most restricted of the bots by design:
- Equities only — individual stocks passing the momentum screen
- No options — no calls, no puts, no spreads
- No leverage — no leveraged ETFs
- No shorts or inverse positions — the strategy is long-only
Keeping the toolkit narrow is intentional. A momentum system lives and dies by its exit discipline, and adding instruments like options introduces decision variables that have nothing to do with the signal.
The regime reflects overall market breadth and trend health — how hospitable the environment is to a long-only momentum strategy:
| Regime | What it means | Capital deployed |
|---|---|---|
| RISK ON | Strong breadth, most sectors trending higher — ideal momentum environment | Highest |
| SELECTIVE | Mixed signals, leadership is narrow — be picky | Moderate |
| RISK OFF | Market weakening, trends breaking down — no new entries, manage existing | Reduced |
| CASH | Hostile conditions — exit everything, preserve capital | 0% |
Exit discipline is what separates this strategy from hope-and-hold. Every position is subject to four hard exit rules:
- Hard stop — position is cut immediately if drawdown hits a fixed percentage level
- Trend break on volume — if the stock breaks its trend on elevated volume, exit regardless of P&L
- Relative strength failure — if the stock meaningfully underperforms the S&P 500 over a short window, exit
- Dead money rule — if a position is flat after a set number of days, exit and redeploy the capital elsewhere; opportunity cost matters
On the winning side, trailing stops activate once a position reaches a set gain, locking in gains as the stock extends. Exact thresholds are proprietary and may change over time.
Why so strict? Momentum strategies work because they cut losers fast and let a handful of big winners pay for the losers. Loosening any of these rules compounds into poor performance over time.
Momentum Quant runs during the trading day, applying its rulebook to entries and exits and monitoring open positions against its exit rules. The exact schedule is proprietary.
Yes. Like the other bots, Momentum Quant continuously learns from its track record — grading the entry, the exit, and the overall setup of every closed position, then identifying patterns across trades.
Because the rules are mechanical, learning here is less about changing the rulebook and more about refining which setups inside the rulebook have historically produced the best outcomes — industry groups, ranking thresholds, volume filters, and similar parameters.
No. All four portfolios are simulated with $100,000 in paper money each. The AI makes real decisions using real market data — the same data any institutional trader would look at — but no actual capital is at risk. This is educational and entertainment content, not a live trading service.
StreetAlpha is not a registered investment adviser. All information is for educational purposes only. Past simulated performance does not indicate future results. Not financial advice.
Swing Trader is the most active — it trades equities, ETFs, and options, and holds positions for 1–10 days. It's built to react to changing conditions quickly.
Value Investor is the most patient — it only buys equities it would be comfortable holding for weeks to months. It ignores short-term noise entirely and focuses on business quality and valuation.
Earnings Trader is the most specialized — it only operates around earnings events and never holds overnight past the report. Its thesis is built around earnings: predicting the post-earnings market reaction, or capturing the run-up in implied volatility and exiting before the report.
Momentum Quant is the most mechanical — it follows a strict rules-based checklist with no discretion. It buys strength, cuts weakness fast, and lets winners run. Price is the only signal that matters.
No. Each persona has its own portfolio, its own decision history, its own lessons learned, and its own $100,000 in capital. They operate completely independently. Swing Trader's lessons don't influence Value Investor's decisions, and vice versa. This keeps each strategy pure and makes it easier to evaluate their performance in isolation.
The bots go dormant during the trading week's off-hours — no decision runs fire and no capital is deployed. Outside of market hours, the bots use the downtime for review and self-improvement work that prepares them for the next trading day.
P&L is calculated based on simulated entry and exit prices at the time the AI records each trade. The portfolio value reflects all open positions marked to current market prices plus any cash held. Figures do not account for bid-ask spread, slippage, taxes, or commissions — these are paper portfolios. Returns are shown gross, before any hypothetical fees.
Yes. Each bot reviews its closed positions, winners and losers alike, and carries what it learns into future decisions. Those lessons build up over time as the bot trades. They are proprietary and aren't shown on the dashboards, but you can see their effect in the decisions and reasoning each bot logs.
Each dashboard shows a performance chart indexed to 100 at inception — so all lines start at the same point and you can directly compare growth. The portfolio (orange line) is measured against three benchmarks: SPY (S&P 500), QQQ (Nasdaq 100), and DIA (Dow Jones). The chart subtitle shows the running outperformance or underperformance versus SPY.