How StockTalk signals are calculated
The StockTalk signals tool analyses each JSE-listed stock against eight technical factors and combines them into a single verdict: BUY ZONE, WATCH, HOLD, or AVOID. This page documents exactly what each factor measures, how they are scored, and what the final verdict means. If you are using a signal to inform a trade decision, this is the page to read first.
1. Support proximity
The model finds the nearest support level at or below the current price — a level where the stock has historically stopped falling. This is scored as a threshold, not a sliding scale. If the price is sitting no more than 3% above that support, the score gains 18 points, the largest single contribution in the model. If the price has closed below the support level, it loses 22 points instead. Anything in between — more than 3% above support — scores nothing either way.
The model also looks up. If the nearest resistance level is within 3% overhead, the score loses 8 points, because a stock running into resistance has less room ahead of it. A smaller bonus of 4 points is added when the price is within 1.5% of a psychological round number (R1, R5, R10, R50 or R100), where buyers and sellers tend to cluster.
2. RSI 14 (momentum)
The 14-day Relative Strength Index measures how fast the price has been moving. The model reads it in three bands, and only two of them move the score. Below 30 is oversold — potentially cheap — and adds 14 points. Above 70 is overbought — potentially stretched — and subtracts 16 points. Between 30 and 70 the RSI is treated as neutral and changes the score by nothing at all.
Overbought does more than cost points: it caps the verdict. A stock with an RSI above 70 can only ever come out as HOLD or AVOID, however well it scores on everything else. See section 8.
3. vs SMA 50 (short-term trend)
The 50-day simple moving average reflects the short-term price trend. The model adds 8 points when the price is at or above the SMA 50 and no more than 4% above it — holding the trend without being stretched. Trading below the SMA 50 earns nothing, and so does trading more than 4% above it; neither is penalised directly.
A further 8 points are added for double support: when the nearest support level sits within 2% of the SMA 50, two independent reasons to expect a bounce line up at the same price.
4. vs SMA 200 (long-term trend)
The 200-day SMA shows the long-term health of the stock, and the model treats it as a straight yes or no: at or above it adds 6 points, below it subtracts 4 points. There is no “close enough” band — a price a cent below the 200-day average is scored the same as one far below it. A stock in a structural downtrend therefore needs stronger signals elsewhere to score well, but it is not ruled out.
5. Volume
Volume is used for one thing only: confirming a breakout. If the price is trading at or above the nearest resistance level and the stock has actually traded that day, the score gains 6 points. A move through resistance on no volume gets nothing.
The model does not compare today's volume against a 30-day average. That average is not stored alongside the other indicators, so rather than estimate it we report the “volume vs 30-day average” figure as unavailable.
6. Catalyst presence
A near-term catalyst adds 6 points: results due within the next 14 days, or a director dealing reported in the last 7 days. The points are a small nudge, but the presence of a catalyst also changes which verdict a given score maps to, and it is the only way a stock can be labelled WATCH — see section 8.
One practical caveat: the screener and comparison tables compute a faster, catalyst-free version of the same signal so they can rank the whole universe in one query. The catalyst-aware verdict is the one shown on an individual stock.
7. Stop-loss risk
Stop-loss risk measures how far the nearest support level is below the current price, as a percentage. A tight gap means a stop-loss order can be placed close to the entry price, limiting the loss if you are wrong. A wide gap means more room to fall before anything catches it. This band is reported alongside the score; it does not feed into the score.
- Low risk: support is within 4% of the current price
- Moderate risk: support is 4%–8% away
- Elevated risk: support is more than 8% away, or no support level could be detected at all
If the price has already closed below its nearest support, the verdict is AVOID outright, whatever this band says.
8. Final verdict
Every stock starts on 50. The factors above add and subtract from there, and the result is capped to the 0–100 range. Turning that score into a verdict is not a simple set of bands — two things can override the score outright, and the remaining ladder depends on whether the stock has a catalyst.
Two overrides, checked first
- Overbought. RSI above 70 means the stock can only be HOLD (score 45 or more) or AVOID (below 45). Never BUY ZONE, never WATCH.
- Support broken. If the price has closed below its nearest support level, the verdict is AVOID even on a high score.
Otherwise, with no catalyst present
- BUY ZONE: score 58 or more
- HOLD: score 46–57 — in trend, but no clear entry
- AVOID: score below 46
WATCH is unreachable on this path. Without a catalyst a stock is never labelled WATCH.
Otherwise, with a catalyst present
- BUY ZONE: score 66 or more and the price within 3% of support (or no support level detected)
- WATCH: score 52 or more, but short of that bar
- HOLD: score 46–51
- AVOID: score below 46
A catalyst therefore makes the model more cautious about calling an entry, not less. A stock on 58 with no catalyst is a BUY ZONE; the same stock with results due next week is held at WATCH until it clears 66 with the price sitting close to support. That is deliberate — a catalyst is a known unknown, and the model would rather flag it for attention than call an entry ahead of it.
This is a purely technical model. It does not consider company fundamentals, news sentiment, or macroeconomic factors, and the thresholds above are judgement calls, not the output of a backtest. Always combine these signals with your own research.
Questions about the methodology, or a specific stock where the signal looks wrong? Email support@stocktalk.co.za and we will investigate. We treat anomaly reports as bugs first, opinion second.