A signal is the app's best guess about whether a stock is likely to go up or down — like a weather forecast, but for stock prices. It's built by checking 6 well-known patterns in the price history and combining them into a single score.

Each pattern is called an indicator. It gives the stock a small score from −2 (bad sign) to +2 (good sign). We add them up. Positive total = lean toward buying; negative = lean toward selling; near zero = unclear, so wait.

The 6 indicators (plain English)

• Trend (50 & 200-day averages) — We smooth the last 50 days of closing prices and the last 200 days into two lines. If the 50-day line is above the 200-day line, the stock is in a long-term uptrend. The opposite is a long-term downtrend.
• RSI — Relative Strength Index — A 0-to-100 speedometer of how fast the stock has been rising vs falling. Below 30 = it fell too fast, might bounce back. Above 70 = rose too fast, might cool off. 50 is neutral.
• MACD — Compares a short moving average to a longer one. When the short one starts rising faster than the long one, that's often the start of an uptrend. When it falls faster, the start of a downtrend.
• Bollinger Bands — Imagine a rubber band around the recent price. When price stretches to the upper edge, it's "overextended up" and often pulls back. Lower edge = overextended down, might bounce.
• Stochastic — Similar to RSI but more sensitive to the last 2 weeks of prices. Good for catching short-term reversals. Above 80 = risky territory; below 20 = potential bounce.
• ADX — Average Directional Index — Measures whether a trend is strong or whether the stock is just drifting sideways. Doesn't tell you direction — just whether there's real momentum to follow.

The 5 signal levels

• STRONG BUY All-or-most indicators agree. Historically right ~70% of the time. Act with confidence.
• BUY Indicators lean positive. Take a smaller position — add more if it gets stronger.
• WATCH Indicators are mixed. No action — wait for clearer signal. Patience usually beats impatience.
• TRIM Weakening — consider taking some profit off the table if you own it.
• EXIT Indicators agree it's time out. Sell if you own it.

Why signals get "downgraded" in a bear market

If the overall trend of a stock is down (50-day average below 200-day average), we automatically downgrade any BUY signal to WATCH. Trading against the dominant trend is hard — better to wait for the tide to turn.

An honest limitation

Signals are wrong sometimes. No algorithm predicts the market perfectly. Even the best professional quant funds win about 55–60% of their trades. Treat signals as one voice in a decision — use your own judgment too. The Research tab shows how our signals perform against a decade of real data so you can see the track record.

Not financial advice. Educational only. You are responsible for your own money.

A backtest replays a trading strategy on historical price data to estimate how it would have performed. Use it as a sanity check — not a crystal ball.

What the simulation does

• Walks day-by-day — starts at the oldest bar of the chosen period and moves forward one day at a time, never peeking at future data (no look-ahead bias).
• Computes the signal — on each day, calculates the composite score using only data available up to that day.
• Entry logic — if score ≥ Buy threshold and currently flat → buy shares at the next bar's open (realistic — you can't trade at a close you haven't seen).
• Exit logic — if score ≤ Sell threshold while long → sell all at next bar's open. If a trailing stop is enabled and price falls X% below the highest close since entry → forced exit.
• Costs — every buy and sell deducts the transaction cost (bps) from the trade value. Default 5 bps ≈ retail commission + slippage.
• Benchmark — a simple buy-and-hold of the same ticker runs in parallel, with costs applied to its one entry and one exit.

How to read the results

• Strategy return — what $10k (or your starting cash) would have become under the strategy.
• Buy & hold — what $10k would have become just owning the ticker for the whole period. If strategy < buy-hold, the rule didn't help for this ticker/window.
• Trades — round-trip count (buy + sell = 1 trade). High trade count with poor win rate = noise, not edge.
• Win rate — % of closed trades that were profitable. >50% is OK; higher is better; but a 40% win rate with big wins can still outperform a 60% win rate with tiny wins.
• Equity curve — the blue line is strategy value over time; dashed gray is buy-and-hold. Smoother is better — jagged drops = drawdowns.

What the settings do

• Buy / Sell thresholds — tighter thresholds (higher buy / lower sell) = fewer, higher-conviction trades. Looser = more trades, more noise.
• Transaction cost — 5 bps is realistic for most US retail brokers. Test with 0 bps to see theoretical upside; test with 20 bps to stress-test.
• Trailing stop — research shows tight stops (<10%) hurt returns in most cases; 15-20% is a reasonable protection without bleeding profits. Off by default.
• Period — longer = more statistical weight. 5-10y includes multiple market regimes. "max" shows decades but becomes less relevant to today's market structure.

Limitations & caveats

Past performance is not indicative of future results. Backtests can be overfit — a rule tuned to look good on old data may fail forward. To guard against this, the Research tab splits history into in-sample and out-of-sample windows. If a rule performs similarly in both, it's more trustworthy.

Backtests also ignore: market impact on large orders, tax drag (capital gains), dividends reinvested separately from price, short-selling, and survivorship bias (dead tickers dropped from Yahoo's feed). Treat results as a rough signal, not precision.

Two kinds of positions

• SESSION — opened through this app via "Mark purchased" on a buy signal. Counts against your session budget.
• MANUAL — pre-existing holdings you imported to track P/L. Does not touch the session budget.

Click either badge on a position row to reclassify.

The session budget

The 💼 budget pill in the top bar tracks how much of your planned capital you've deployed this session. It turns amber at 80%+, red over 100%. The goal: keep you from over-trading.

Sizing: STRONG BUY uses % of total budget, BUY uses %. Adjust both in Settings.

P/L math

• Unrealized P/L — `(current price - cost basis) × shares`. What you'd make/lose if you sold right now.
• Realized P/L — locked in by recorded sales. Shows up in the trade log + "Realized P/L · all time" panel.
• Day change — `(current - previous close) × shares`. Today's movement only.
• Cost basis method — Average (blended), FIFO (oldest lots first), or LIFO (newest first). Choose in Settings → Defaults.

Price data + lag

Prices come from Yahoo Finance — usually within 15 min of your broker's feed but not always identical intraday. For end-of-day reconciliation (after 4:15 PM ET), numbers should match. If you want live quotes, add an Alpha Vantage key in Settings.

The Research tab runs every candidate rule the app knows against a diversified universe of US equities — so you can see which rules have genuine edge vs which are noise.

What gets tested

• 12 candidate rules — including our live composite, golden cross, MACD trend, breakout, dip-buying, ensemble voting, regime switching, and continuous sizing.
• Universe sizes — small (~37 tickers), medium (~95), large (~130). Larger = more statistical weight but slower.
• Period — 2y / 5y / 10y. Longer includes more market regimes.

In-sample vs out-of-sample

We split history into the first 60% (in-sample) and last 40% (out-of-sample). A rule tuned on in-sample that performs similarly on out-of-sample is not overfit — we can trust it. If out-of-sample is much worse, the rule's edge was an illusion.

Reading the table

• Alpha — Return minus buy-and-hold. Positive = rule adds value. Negative = rule subtracts.
• Win% — per-trade win rate. Composite's 65-70% is the best in the family.
• Sharpe — risk-adjusted return. 0.4+ is respectable for single-name strategies.
• MaxDD — worst peak-to-trough decline. Smaller = smoother ride.

Alerts are rules you define that fire whenever data refreshes and your condition is met. They trigger a browser notification (if you've granted permission) and log to the recent-triggers list.

Rule types

• Signal-based — "Signal is BUY", "Score above 5", etc. Fires when a ticker's composite crosses your threshold.
• Price-based — "Price above $X", "Price below $Y". Simple stop/take-profit triggers.
• Indicator-based — "RSI above 70" (overbought), "RSI below 30" (oversold).
• Position-based — "Down X% from my cost basis", "Up $Y from cost". Only works for tickers you own. Set them from the Portfolio tab.

Evaluation

Rules run whenever the app refreshes (ribbon, watchlist, home scan, portfolio). The same alert won't fire twice for the same condition — it only re-fires when the state changes and crosses back in.

Notifications

The first time you click "enable notifications" on the Alerts page, your browser will ask for permission. Accept to get OS-level popups when rules fire. Log-only alerts work without permission.

Rebalancing is keeping your portfolio's sector mix close to your target. If one sector rallies, your allocation drifts — you take profits there and rotate into underweights.

The rebalance tool

• Set targets — click a target % cell for each sector. Aim for your preferred mix (e.g., 30% Tech, 20% Healthcare, 20% Financials, etc.).
• Read Δ — shows the dollar amount to buy or sell in each sector to hit target.
• Execute — go to your broker, make the trades, then use "sell" or "mark purchased" in the app to log them.

Classic mixes for reference

• Growth — 40% Tech, 20% Comm Services, 15% Consumer Discretionary, 15% Healthcare, 10% Financials.
• Balanced 60/40-style equity — roughly equal-weight across all 11 sectors.
• Defensive — 25% Consumer Staples, 20% Healthcare, 15% Utilities, 15% Real Estate, 10% each of Tech / Industrials / Energy.