Long-term trend

Most people who own shares already do a rough version of this: hold while things are going well, get out when they are not. The trouble is deciding what 'going well' means on any given day. This strategy decides it the same way every time.

It uses two averages of the price: one over the last 50 trading days, one over the last 200. When the shorter average climbs above the longer one, the recent past is better than the distant past, and the company is bought. When it falls back below, it is sold. That is the whole rule.

Trends in large companies tend to last months, not days. A rule that only reacts when the 50-day average crosses the 200-day ignores every wobble in between, which is why it trades so rarely. That same slowness means it is late at both ends: it buys some way up from the bottom and sells some way down from the top. What it captures is the long middle.

A stop sits well below the entry in case the new trend fails at once, and each position is sized so that hitting it costs a fixed small fraction of the account. Ten companies, never more than 15% of the account in one.

Illustrative. ₹1,00,000 account, balanced version, a stock at ₹1,000 whose 50-day average has just crossed above its 200-day.

Entry

₹1,000, the day of the cross

Stop

₹850, 15% below

Risk budget

1% of the account, ₹1,000

Position

about 6 shares, ₹6,000, so the stop costs ₹900

Exit seven months later

₹1,180, the day the 50-day crosses back below

Result

₹1,080 before charges, about 1.1% of the account

A single position is small because the stop is wide and the risk budget is fixed. The return comes from holding several of these through a long rise, not from any one of them. In a sideways year the same rule produces a few small losses and nothing else.

These are the real rules, in the order the strategy checks them. Nothing here needs you to be watching.

1. 1

   Sells when the trend turns

   The 50-day average crossing back below the 200-day is the exit, and it will usually be some way below the top.

2. 2

   Stops if the new trend fails at once

   Twelve to eighteen percent below the entry, by version. Wide, because a trend needs room, and sized so that stop costs the same small fraction of the account every time.

3. 3

   Never more than 15% of the account in one company

   However long the trend has lasted.

4. 4

   Holds for as long as it takes

   This is a delivery strategy. Months are normal. A trade that lasts a year is the point.

Adjustments, and why it does not make them

Investors adjust this kind of rule by taking profits on the way up or adding on dips. Both feel sensible and both shorten the long middle the strategy exists to capture.

This robot does neither. It holds until the cross, or the stop, and the same company can only be bought again on the next cross.

Each strategy comes as conservative, balanced and aggressive. They share the idea and differ in how much they risk once in.

The only difference between the versions is how much is at stake per trade. The signal is the same.

• Long only, and slow to leave. In a crash the exit arrives after a chunk of the fall; the stop limits how large a chunk.
• Overnight and weekend risk, as with any shares held for months.
• Few trades. Over five years each company produces a handful, so the record is short and a backtest's numbers carry wide error bars.

EMA

An average of recent prices that leans towards the latest ones. Price above its EMA is drifting up; below, drifting down.

ATR

The typical size of one bar's move. Stops in these strategies are set a few ATRs away so ordinary noise does not trigger them.

Delivery (CNC)

Shares bought to hold overnight or longer, paid for in full. No shorting.