I'm also a $25/unit player here. Small time.
Fresh it really depends on what your bankroll is and your tolerance for risk. Typically, most people consider a unit of play to equal 1% of their entire bankroll. So for me, if I ever go down $2,500 I am busto and calling it a year.
I am a hobby player and a fairly risk averse player myself, so I try to limit my unit play per game. Of course I want to make money, but the entertainment factor is the #1 reason why I bet. I will usually go 1u or 1.25u on games that I like, and up to 3u on games that I REALLY like. Other people have different systems and will go 1-5u plays (obviously based on how much they like that particular bet) but will go 6-10u plays on their "lock" type of plays.
My only exception to my rule (and it is a big exception) is on MMA betting because I often bet big favorites. In these situations I find myself risking upwards of 4-6 units to win 1-2 units. With MMA bets I work with a partner and we try to breakdown and assess future probabilities on outcome.
quick example:
Lets take this weekends Maynard v. Edgar UFC title fight.
I will try to breakdown each fighters different paths to victory and then break down the probability of each of those paths chance of occuring. We then compare our assessed probability against the current betting lines to see if there is any +EV on either side. Obviously this is an imperfect science (since the key data "probability of certain outcomes" is somewhat arbitrarily decided by me), but it helps give both me and my partner some pseudo-science based reasoning behind our picks.
So say after doing all our analysis we think that its gonna be real tight and Maynard has a 48% chance of victory and his current betting odds are at +125. +125 implies that Maynard has a 44.45% chance of victory (100/225). Therefore if we truly believe our analysis and probability is correct at 48% chance of victory, there would be slight value at that price.
Obviously the more perceived value, the bigger the unit bet size.
This past weekends card had a fight that my boy and I thought was a huge value play. Anthony Rumble Johnson opened at -185 against Charlie Brennamen. We believed that Rumble legitimately had a 80% chance or better to win that fight whereas -185 would put it closer to 65%. We bet big units on that fight, and thankfully we were correct with our assessment (we got crushed on another fight, but that is not relevant to this discussion
)