Magic Missile,
if you have a single target and are lucky or an evoker
"If you are lucky, you will do more damage" is true of any spell with damage dice. But upcast spells typically have multiple damage dice, and as a result of the law of averages the more dice you have, the more likely your damage will be close to the mean.
But, RAW, magic missile has only a single damage die; you roll just once and apply the roll to every missile. And it is a small die, just a d4. That means that every time you cast the spell, you have a 25% chance of achieving maximum damage, regardless of the number of missiles. Without a feature that grants it (such as overchannel), the chance of maximum damage of other spells becomes vanishingly small the more dice you add with upcasting. Thus upcasting a typical damage spell is more likely to yield a total closer to the average, which is not true of magic missile.
We can compare magic missile with other spells chosen specifically for their high output of damage per level. We should also remember that very few things resist force damage, and that magic missile damage scales well since it does not require an attack roll and does not permit a save.
The results show that for a single target, magic missile is competitive with the best damage spells of each level (inflict wounds, fireball, blight, and disintegrate). Just as higher level spells begin to outstrip it in damage, the empowered evocation ability, if you have access to it, provides significant increases in damage. If for some reason you are limited in your spell choice, magic missile is a robust selection, given that it is among the most effective spells at slots from first to eighth level.
Magic missile cast with a first level slot gets you 3 missiles, with an average damage of 10.5, but a 25% chance of the maximum damage of 15. In comparison, inflict wounds cast with a first level slot has an average damage of 16.5, but a 25% chance of damage around 23.25 (assuming you are in the upper quartile of damage, with results weighted proportionally by the chance of each roll). With an estimated 2/3rds chance of hitting your target, you could expect a lucky inflict wounds to do 15.5 damage.
'Lucky' magic missile: 15
'Lucky' inflict wounds: 15.5
Magic missile cast with a second level slot gets you 4 missiles, with an average damage of 14, but a 25% chance of the maximum damage of 20. Inflict wounds cast with a second level slot has an average damage of 22, but a 25% chance of damage around 29. With a chance of hitting your target, you could expect a lucky inflict wounds to do 19.3 damage.
'Lucky' inflict wounds: 19.3
'Lucky' magic missile: 20
Magic missile cast with a third level slot gets you 5 missiles, with an average damage of 17.5, but a 25% chance of the maximum damage of 25. Fireball cast with a third level slot has an average damage of 28, but a 25% chance of damage around 341. Taking 83% of the damage to account for the chance of the target making its save, you could expect a lucky fireball to do 28.48 damage.
'Lucky' magic missile: 25
'Lucky' fireball: 28.5
Magic missile cast with a fourth level slot gets you 6 missiles, with an average damage of 21, but a 25% chance of the maximum damage of 30. Blight cast with a fourth level slot has an average damage of 36, but a 25% chance of damage around 441. Taking account for the chance of the target making its save, you could expect a lucky blight to do 36.8 damage.
'Lucky' magic missile: 30
'Lucky' blight: 36.8
Magic missile cast with a fifth level slot gets you 7 missiles, with an average damage of 24.5, but a 25% chance of the maximum damage of 35. If you are casting 5th level slots, you might be 10th level or higher, and if you are an evoker, you can use your empowered evocation to add +5 to each missile (assuming 20 Int) for a further bonus of 35 damage and a maximum of 70. From here on up, you don't really have to rely on the 25% chance of maximum damage damage, since the contribution of the empowered evocation is so significant. Blight cast with a fifth level slot has an average damage of 40.5, but a 25% chance of damage around 49.51. Taking 83% of the damage to account for the chance of the target making its save, you could expect a lucky blight to do 41 damage.
Average magic missile: 24.5
Average blight: 33.6
'Lucky' magic missile: 35
'Lucky' blight: 41
Average empowered magic missile: 59.5
'Lucky' empowered magical missile: 70
Magic missile cast with a seventh level slot gets you 9 missiles, with an average damage of 31.5, but a 25% chance of the maximum damage of 45 and a maximum empowered damage of 90. Disintegrate cast with a seventh level slot has an average damage of 87.5, but a 25% chance of damage around 93 1.
Average magic missile: 31.5
'Lucky' magic missile: 45
Average empowered magic missile: 76.5
Average disintegrate: 87.5 with failed save (no damage if save made)
'Lucky' empowered magical missile: 90
'Lucky' disintegrate: 93 with failed save (no damage if save made)
Magic missile cast with a eighth level slot gets you 10 missiles, with an average damage of 35, but a 25% chance of the maximum damage of 50 and a maximum empowered damage of 100. Disintegrate cast with a eighth level slot has an average damage of 96, but a 25% chance of damage around 105 1.
Average magic missile: 35
'Lucky' magic missile: 50
Average empowered magic missile: 85
Average disintegrate: 96 with failed save (no damage if save made)
'Lucky' empowered magical missile: 100
'Lucky' disintegrate: 105 with failed save (no damage if save made)
At a ninth level slot Meteor swarm, with an average damage of 140, and a 25% chance of damage around 154, is pretty much untouchable by other single-round damage spells.
Arcane hand, in a long fight
For a fifth level slot, the 4d8 force damage of a Clenched Fist to a single target is less than the single-round damage of every 5th level Area of Effect spell. Clearly it is not a good choice, especially for a typical three round fight. But when upcast, it gains 2d8 damage for each spell slot, which is good scaling. In a prolonged, tactical fight that might see it on the board for its full duration, the total possible damage is significant. Other damaging spells typically grant saves, which means something like a target's magic resistance dramatically drops expected damage, whereas the Clenched Fist need only hit. Likewise force damage is one of the least-resisted damage types. After the initial cast, it uses a bonus action to strike, which means a non-Concentration caster like an evoker can use it to supplement whatever their main cast is each turn.
Upcast with 6th level slot, it hits for 6d8 (average 27). Assuming two-thirds of attacks hit (but not adjusting for the significant increase of damage on a crit), that is 180 over ten rounds (or 213 with enhanced evocation) with a single 6th level slot.
With a 7th level slot, 8d8+5 is 273 over ten rounds of attack rolls.
With an 8th level slot, 10d8+5 is 333 over ten rounds of attack rolls.
With a 9th level slot, 12d8+5 is 393 over ten rounds of attack rolls.
When I DM'd a 20th level party's climactic campaign-ending fight against Tiamat, the evoker initially wondered how he was going to contribute, given Her high saves, magic resistance, legendary resistance, and elemental immunities. Eventually he chose a strategy with arcane hand as part of both defense (sheltering behind a 7th level wall of force and assailing the Mother of All Dragons with a highly upcast Clenched Fist) and offense (attacking in melee with Tenser's Transformation while throwing in an extra 59hp Clenched Fist punch each round as a bonus action).
1 Upper quartile damage estimate should be used for comparative purposes only. Anydice doesn't display probabilities of less than 1/100th of a percent, and I didn't take the time to calculate these out exactly for the increasingly long tails of the high dice spells.