Chapter 1 of 7
How the suggestions work
What the algorithm does, step by step, with nothing hidden.
Step 1: Decide which stocks it may suggest
Every trading day after the market closes, it looks at every security listed on NEPSE (432 today). It keeps only ordinary shares and mutual funds that are actively trading, have at least 120 trading days of history, traded on at least half of recent days, and trade enough each day: a median of Rs 3 lakh a day for shares and Rs 1 lakh for mutual funds over the last 120 sessions. Promoter shares are left out. Today 272 pass.
Step 2: Score each one on ten simple measures
Momentum (6- and 12-month return), trend (price against its 200-day average), low volatility, shallow past falls, long-run growth (3 to 5 years), consistency (how often a 12-month period ended positive), value (how far it sits below its 52-week high and 3-year average), liquidity, company size, and how much it moves with the market (beta). On each measure a stock gets a rank from 0 to 1 against the other eligible stocks.
Step 3: Weigh the measures by your time horizon
For short horizons (6 months or less), momentum, liquidity and low volatility count most. For long horizons (5 years or more), long-run growth, consistency and value count most. In between, the weights blend smoothly. For a 2-year plan the biggest weights are liquidity (17%), momentum (16%) and low volatility (14%).
Step 4: Build a plan you can actually buy
It picks up to 8 holdings, about one for every Rs 8,000 (smaller amounts get fewer), with at most two from one sector and a limit on mutual funds. Higher-scoring, calmer stocks get a bigger share, but no stock gets more than 35%. It prefers stocks you can buy within 5 trading days without taking more than 10% of their normal daily turnover. It then buys whole 10-share lots, counting broker commission, the SEBON fee and Rs 5 per order, and keeps the leftover cash for your next contribution.
Step 5: Each time you add money
It buys one lot of the holding that is furthest below its target share, if the cash covers it. If not, the cash waits for the next contribution.
Step 6: Show a range, not a promise
To show what might happen, it replays this exact mix’s own monthly returns from the last 10 years, in random blocks of a few months, 2,000 times. It takes off buying and selling costs and capital gains tax (5% if held a year or less, 3.75% if longer). You see a weak result (worst 10%), a typical one (the middle) and a strong one (best 10%), and how often the plan ended below the money put in.
1 / 7