Generate a fictitious stock option price variation - php

Question

What I do
I am making graph of fictitious stock options.
The price is updated each second, with this function
function stockVariation($price,$max_up,$max_down)
{
// Price > 1
if($price > 1)
{
// Calculate
$ratio=(mt_rand(0,$max_up/2)-mt_rand(0,$max_down/2))/1000;
$price+=$ratio;
}
// Price <=1 (we don't want 0 or negative price...)
else
$price+=mt_rand(1,$max_up)/1000;
return round($price,3);
}
I use a max_up and max_down values (from 10 to 100) to make the price change progressively and simulate some volatility.
For example, with max_up : 40 and max_down : 45, the price will progressively go down.
My question
But the problem, is that prices generated are too much volatile, even if max_up = max_down.
The result is "non-natural". (for example +10 points in one day for a base price of 15,000).
Result of price evolution per hour in 24 hour
Perhaps making round($price,4) and divisions by 10 000 instead of 1 000, will be better ?
If anyone have an idea or an advice to generate "natural" prices evolution, thanks in advance.

Answer 1

There are 86400 seconds in a day, so you'll need to divide by a much larger number. And rather than adding and subtracting, you may want to multiply the current price by a factor that's slightly larger or smaller than 1. That would simulate a percentage increase or decrease, rather than an absolute gain or loss.
function stockVariation($price, $max_up, $max_down)
{
// Convert up/down to fractions of the current price.
// These will be very small positive numbers.
$random_up = mt_rand(0, $max_up) / $price;
$random_down = mt_rand(0, $max_down) / $price;
// Increase the price based on $max_up and decrease based on $max_down.
// This calculates the daily change that would result, which is slightly
// larger or smaller than 1.
$daily_change = (1 + $random_up) / (1 + $random_down);
// Since we're calling this function every second, we need to convert
// from change-per-day to change-per-second. This will make only a
// tiny change to $price.
$price = $price * $daily_change / 86400;
return round($price, 3);
}

Answer 2

Building upon the idea, you could use an actual volatility number. If you want e.g. a volatility of 35%/year, you can find the volatility per second. In pseudocode:
vol_yr = 0.35
vol_month = vol_yr * sqrt(1.0/12)
vol_second = vol_yr * sqrt(1.0/(252 * 86400)) # or 365 * 86400
Then, every second, you "flip a coin" and either multiply or divide current stock price by (1 + vol_second). This is the principle of how binomial trees are created to evaluate exotic stock options.