5 minutes is a really difficult distance - like the mile in a run: not a sprint but not a long slow thing either.
This study says it is better to go off fast.....
Effects of starting strategy on 5-min cycling time-trial performance.
The importance of pacing for middle-distance performance is well recognized, yet previous research has produced equivocal results. Twenty-six trained male cyclists (VO(2peak) 62.8 +/- 5.9 ml . kg(-1) . min(-1); maximal aerobic power output 340 +/- 43 W; mean +/- s) performed three cycling time-trials where the total external work (102.7 +/- 13.7 kJ) for each trial was identical to the best of two 5-min habituation trials. Markers of aerobic and anaerobic metabolism were assessed in 12 participants. Power output during the first quarter of the time-trials was fixed to control external mechanical work done (25.7 +/- 3.4 kJ) and induce fast-, even-, and slow-starting strategies (60, 75, and 90 s, respectively). Finishing times for the fast-start time-trial (4:53 +/- 0:11 min:s) were shorter than for the even-start (5:04 +/- 0:11 min:s; 95% CI = 5 to 18 s, effect size = 0.65, P < 0.001) and slow-start time-trial (5:09 +/- 0:11 min:s; 95% CI = 7 to 24 s, effect size = 1.00, P < 0.001). Mean VO(2) during the fast-start trials (4.31 +/- 0.51 litres . min(-1)) was 0.18 +/- 0.19 litres . min(-1) (95% CI = 0.07 to 0.30 litres . min(-1), effect size = 0.94, P = 0.003) higher than the even- and 0.18 +/- 0.20 litres . min(-1) (95% CI = 0.5 to 0.30 litres . min(-1), effect size = 0.86, P = 0.007) higher than the slow-start time-trial. Oxygen deficit was greatest during the first quarter of the fast-start trial but was lower than the even- and slow-start trials during the second quarter of the trial. Blood lactate and pH were similar between the three trials. In conclusion, performance during a 5-min cycling time-trial was improved with the adoption of a fast- rather than an even- or slow-starting strategy.