Technical Analysis on the PSX: Using Moving Averages, RSI and MACD Without Fooling Yourself

Technical analysis has a strange status among Pakistani retail investors. Half the market treats it as astrology, the other half as a money machine. Both halves are wrong in the same way: they think indicators predict. They do not. A moving average, an RSI value, a MACD histogram — these are descriptive statistics of past prices, nothing more. We have already written about what these tools cannot tell you; read that first, because the limits matter more than the mechanics. This piece is the companion: how to actually use the three most common indicators, in the right order, on a market with its own peculiar distortions.

A moving average defines the trend. That is its entire job

The 50-day moving average is the average closing price of the last 50 sessions. Each day the oldest close drops out and the newest drops in. The 200-day is the same arithmetic over roughly a trading year. Nothing more sophisticated is happening.

Its use is definitional. "Price above a rising 200-day average" is not a prediction that the stock will go up — it is a definition of an uptrend, agreed on by enough participants to be a shared language. Trend is otherwise a matter of opinion; the moving average makes it a matter of arithmetic.

Then there is the golden cross — the 50-day crossing above the 200-day — and its mirror, the death cross. Understand what they actually are: the 200-day average contains ten months of history, so by the time the 50-day climbs through it, the rally that caused the cross has already happened. The cross is a lagging label stuck on a move that is months old. It is not useless — as a slow regime filter, "only hold long positions above a rising 200-day" would have kept you out of most of the PSX's grinding bear stretches. But it charges a fee for that protection: you surrender the first leg of every recovery. That trade-off is acceptable. Treating the cross as an entry signal is not.

RSI has one genuinely useful read, and it is not the 70 line

RSI-14 compares the average size of up days against the average size of down days over the last 14 sessions, and maps the result onto a 0–100 scale.

The convention says above 70 is "overbought" and below 30 is "oversold", and this vocabulary has cost retail investors more money than almost any other technical idea. Overbought is a description of the recent past, not a forecast. In a genuine uptrend — think of the KSE-100's strongest rallies, or a re-rating run in a name like SYS or MEBL — RSI sits above 70 for weeks while the price keeps climbing. Selling because a number crossed an arbitrary line means systematically exiting your best positions early and calling it discipline.

The read that earns its place is divergence. Price makes a new high; RSI makes a lower high. Buyers are still lifting the price, but with measurably less force than last time. That is worth noticing — not as a sell instruction, but as a prompt to check everything else: is volume fading too, has the valuation run ahead of earnings, is the sector story exhausted? Divergence also fails, frequently. It just fails less expensively than line-crossing, because it asks a question instead of issuing an order.

MACD: elegant arithmetic that hates sideways markets

MACD is the 12-day exponential moving average minus the 26-day one. A 9-day average of that difference is the signal line, and the histogram is the gap between the two. When the histogram flips positive, short-term momentum has turned up relative to the medium term.

Because it is built entirely out of moving averages, it lags by construction. The bigger problem is what it does when there is no trend to lag. In a sideways market both averages flatten and converge, the histogram oscillates around zero, and every wiggle of noise flips its sign. Each flip looks like a signal. None of them are.

This matters here more than most places, because the PSX spends a great deal of its life sideways. The index went essentially nowhere for long stretches between 2017 and early 2023, while the policy rate climbed to its 22% peak and money hid in T-bills. Run MACD over any of those stretches and you get dozens of crossovers, most of them false, each one costing commission and capital gains tax on the round trip. MACD is a trend-following tool. It pays only in trends, and it bills you for every sideways month.

The order: regime first, momentum second, entry last

No indicator means anything in isolation. The workflow that survives contact with real markets runs in a strict order:

1. Regime. Is there a trend at all? Check ADX — below roughly 20, there is no trend and every trend-following signal is noise — or simply look at the slope of the 200-day average. Flat average, no regime, stop here.
2. Momentum. Only inside a confirmed trend do RSI and MACD earn a reading. Is momentum confirming the trend or diverging from it?
3. Entry and exit. Position size and stop level, decided before you buy, in writing. That discipline is a separate subject — covered properly in our piece on risk management in frontier markets — and it matters more than everything above it.

Most retail traders run this backwards: they see an RSI number, buy, and only then wonder whether there was a trend. Here is what each tool is actually for:

Where the PSX corrupts the inputs

Everything above assumes the price series is honest. On the PSX, two things routinely break that assumption.

Thin volume makes indicators decorative. Hundreds of listings trade a few thousand shares a day. RSI-14 on such a stock is summarising the behaviour of a handful of trades — possibly a single participant testing the order book. The maths still produces a number between 0 and 100; the number describes nothing. Before reading any indicator, check average daily volume, and if a position of your size could not exit in a day or two without moving the price, close the chart. The screener lets you filter by volume precisely so this check comes first.

Circuit-locked closes poison the series. PSX enforces daily price fluctuation limits — as of mid-2026 roughly ±10% for most securities, but verify the current rules in PSX notices, as they change. When a stock locks limit-up or limit-down for consecutive sessions, each close is not a market-clearing price. It is an administrative ceiling with an unfilled queue behind it. Feed a run of locked closes into a moving average or RSI and the indicators read a smooth, powerful trend that never actually traded. RSI pins at an extreme, MACD shows majestic momentum, and none of it reflects a price you could have dealt at. After any circuit-locked episode, indicator values are corrupted until several sessions of free trading rebuild the series.

Ten minutes on a liquid name

Here is the workflow on a stock where it can actually work — say OGDC, or HBL, both liquid enough that the tape reflects a real market.

First, volume: is turnover in line with the stock's normal range, so the prices feeding the indicators were set by genuine two-way trade? Second, regime: is the price above or below the 200-day average, and is that average rising, falling or flat? A flat 200-day ends the technical part of the analysis immediately — whatever you do next is not trend-following. Third, agreement: does the 50-day tell the same story, or is price whipsawing across it? Fourth, RSI — and not the level. The shape. Compare the last two or three price peaks with the RSI peaks beneath them and look for divergence. Fifth, MACD histogram, strictly as a timing refinement inside a trend already confirmed by everything above. And sixth, before any order: the exit, written down — where the stop sits, what position size that stop implies, what would prove the idea wrong.

Notice the proportions. Most of those ten minutes go to deciding whether the indicators deserve to be read at all. That allocation is the skill.

The settings are not magic — and tuning them is worse

Why 14 days for RSI? Because J. Welles Wilder picked 14 when he published the indicator in 1978, for commodity markets, on daily charts, before Pakistan had an electronic exchange. Nothing about the PSX makes 14 correct. The 12-26-9 in MACD is equally an accident of history.

The wrong response to this is the one most people reach for: backtest 9 against 14 against 21 against 25 until one of them "works" on your favourite stock's last three years, then trade it with conviction. A parameter tuned until history looks profitable has learned that particular history, not the market. If your trade appears at RSI-21 and vanishes at RSI-14, you have not found an edge — you have found a coincidence with a parameter attached, and it will not survive out of sample. The defaults are arbitrary, but they are honestly arbitrary: nobody fitted them to flatter your backtest. That is a genuine virtue. It is the same overfitting trap we describe in what technical indicators cannot tell you, and it catches sophisticated people precisely because tuning feels like rigour.

One more thing worth sitting with. On a liquid PSX name, the counterparty to your trade is frequently an institution whose analyst has the same chart, the same RSI, the same MACD. Whatever edge exists in that transaction, it is not the indicator — both sides can see it. The edge, if you have one, is the discipline wrapped around it: reading regime before momentum, refusing the thin stocks, writing the exit before the entry, and leaving the parameters alone. Indicators are a shared language for describing the past. The person who mistakes them for a private window into the future is the person the language was invented to profit from.