"Safe" percent of bankroll for win bets?

[jackad]

I would like to know what you would consider a safe percent of a bankroll to use as flat bets in win only betting. Thanks for your opinions.
Jack

[anotherdave]

Depends on your results. I win about 30% of my win bets. I bet a basically a flat bet, but it is usually about 1.0%-1.5% of my bankroll. ( If I was a long shot player and won 15 or 20% of my bets, I probably would bet 0.5% or less of my bankroll. If I could hit 35-40% I wish! I might bet 3%)

I adjust the bet amount after my bankroll goes up by 25%. I do not adjust it down.

AD

[so.cal.fan]

A man once gave me a win bet plan, which is pretty sound.
You take a set amount you can afford, say $1000.
Bet 5% or $50 bets. Do not decrease them. Stick with $50 bets.
This allows you 20 bets. If you are winning, increase at 5% of profits. If you have a bad meet, you will have made $50 flat bets.
A good meet, you can win a good amount.
The key is to have enough money for betting that it doesn't bother you.
Anotherdave's plan is good. Ideally, you would have at least $5000 betting capital, and break it into 5 $1000 starting stakes.
If you can't hit a good streak with that, you probably need to change your handicapping method.

[GameTheory]

5% is too much unless you're just looking to make a quick strike and don't mind losing. It is not that unusual to see even the favorite lose 14 in a row.

By the way, if you're in a losing streak and losing confidence in your ability, I recommend you just grab a stack of charts and jot down how well the favorites did on any given sample. You'll likely be amazed at the length of some of the losings streaks. Even if you can pick 33%, the up & down swings can be pretty dramatic...

[tanda]

The answer can not be given based on the information submitted.

To determine the amount of bankroll to bet, you need three pieces of information.

1) Win percentage
2) Average odds of winners
3) Risk of "tap-out"

If you have all three components, you can determine the exact percentage to bet. Number 3 is a matter of personal preference. Obviously, numbers 1 and 2 must themselves be verified through extensive testing.

You can also calculate a good approximation with numbers 1 and 3 and without number 2.