Lots of people modify their car by spending money on something
someone else tested. They buy a kit that has 8 or 10 thousand
hours of engineering behind it, bolt it on and hopefully everything
works okay.

If you're going to do something nobody has done before, expect
to either spend a fraction of that time (100-2000 hours) getting
your setup sorted out or expect that it will only operate at 50
to 80% of optimum relative to what you could get with better
tuning. (The other thousands of hours are to make sure that the
kit produces correct results on a wide variety of environments
and installations.)

The dyno is important for tuning because without it you have
no hope of isolating variables. You want to make a motor that
makes a maximum of torque under all throttle and RPM settings.
You don't *really* need an exhaust gas reader but they're
helpful; a motor that is producing a maximum of torque for a
given air flow will be operating at mostly correct air/fuel
mixture so it is somewhat redundant. All proper dynomometers
will have an exhaust gas analyzer as part of the kit.

Anyhow, you don't want to buy a bazillion needles but you
also need to make 2 needles that are identical. The easiest
way to do that is to put notches in your needles. Originally
they were likely made round instead of square or triangular or
oval shaped because it is easier to mass produce them round on
a lathe in a way similar to how keys are reproduced. It is
easier to reproducibly modify them using a file calipers and
a ruler.

All this applies only if you are dinking with carburetors.
If you are using a computerized fuel injection system the
device comes with sensors and such and can magically deal with
things far better than a carburetor.

Most people modify their cars by applying cubic dollars and
faith. Others use cubic dollars and cubic hours. Without some
proper instrumentation, though, much of that is likely thrown
right out the window because the process of trail and error
only works if you know you've made an error...
chris